You are here: UK Essays » Essays » Information Technology » Voice Internet Protocol

Cookie Information

Privacy Information

Voice Internet Protocol

ABSTRACT

Voice over Internet Protocol (VoIP) is an emerging telecommunications technology that is shaping the future of telephony. It is a technology which enable the transmission of voice over the IP networks. It not only carries the voice traffic but also carries video and data traffic across the networks, so it is not erroneous saying that it enables the convergence of data, voice and video traffic.

This project confides with almost all the technicalities of VoIP starting from making a phone call to critical analysation of its voice steams. It takes you to the history and evolution of VoIP and then how it advances, how we see its progress today in enterprise networks and why there is need for the enterprises to implement this robust technology in their infrastructures. From history it will take you to the real working of VoIP. It construes how VoIP process initiates, its deployment models, equipment, and types of protocols that helps voice to propagate over the WAN links. Further it will discuss about the enterprise design, and then this report will talk about WAN design of Frame Relay and Wireless. The report further describes in detail how MQC (Modular QoS CLI) is used to implement QoS in the routers. How it will affect the call quality and what is its need? And what are the current trends of QoS. It also gives the analysis of three important characteristics of voice jitter, packet loss and delay.

CHAPTER NO.1

INTRODUCTION

Ever since Alexander Graham Bell invented telephone in 1876, telephony systems have been a part of our life. Most recently, with the advent of Internet much of our communication is done electronically using the IP network. It has connected the whole world otherwise separated by physical barriers, providing an inexpensive yet quite reliable mean of data transfer and information sharing. Nowadays these two communication means are merging and enterprises are putting their telephone setups over there data carrier setups using a technology called Voice over IP (VoIP).

Voice over Internet Protocol (VoIP) is an emerging technology that is revolutionizing the future of telephony. It uses Internet Protocol (IP) for communications.The aggressive growth of data traffic in enterprise networks require them to implement an infrastructure through which they pass integrated services, voice, data and multimedia. The best solution is to put every thing on IP.So VoIP is the only future for all these robust integrated services.

A new report from analysts group Forrester Research (2005) shows that 36% of North American enterprises and 32% of European enterprises expect to increase their VoIP spend in 2006. It will be several years before enterprises are fully converted to IP Telephony, says Forrester, and the result will be the hybrid including new and legacy environment over the next five years. In another growth analysis research, Infonetics projects that the number of residential and small business enterprises, VoIP users in North America alone will reach 39 million by 2009. Europe is expected to account for 42.5 million by 2009, with the number of users in Asia projected to grow to 49.9 million.Infonetics Research (February 2006).

The reason for this remarkable growth in enterprise networks is using VoIP(packet switching technology) because it reduces the time needed to maintain a connection between two sources, reducing the load on a network and is also not restricted to local networks or wide-area networks(WANs), such as the Internet and most importantly it is cost effective.

1.1 Advantages of VoIP

VoIP has several advantages over traditional circuit switched telephony:

• Users can evade the toll charges levied on calls made over the PSTN.

• Lower cost of equipment.

• Integrates both voice and data functionality.

• Lower bandwidth requirements.

1.2 Problems with VoIP and Need of Quality of Service (QoS)

In spite of these advantages, the biggest challenge to adopting VoIP services is to provide "service quality" and "reliability" equal to existing PSTN circuit-switched networks. IP faces a different challenge than PSTN, such as delivering voice packets in sequential order. When using VoIP, systems transmit IP packets in any order. The receiving VoIP system must reorganize the packets in sequential order and verify no packets are missing i.e. no packet loss. This difference in the way voice packets transmit means VoIP also has to verify and deliver a certain level of Quality of Service (QoS), as well as avoid latency, jitter that can affect VoIP calls differently than calls placed over the PSTN.

1.3 Project Aim

To compare and analyze the performance of VoIP in two different WAN technologies namely (Frame Relay and Wireless) in an enterprise and implementation of Quality of Service(QoS) for the provision of better quality of voice.

1.4 Project Objectives

The project has the following primary and secondary objectives.

1.4.1 Primary Objectives

• To design a network in LAB environment to simulate enterprise network of two different WAN technologies.

• To ensure the connectivity of analog and soft phones over the Frame Relay and Wireless links.

• To design a test bench and implement QoS for voice in an enterprise and typical WAN sites namely (Frame Relay and Wireless).

• To provide a critical comparison of WAN technologies where VOIP can be deployed in an enterprise.

1.4.2 Secondary Objectives

• To study and analyze the behavior of VoIP in other WAN technologies such as Cable, MPLS VPN and Metro Ethernet.

• To critically compare the test results of all the five WAN technologies i.e. (FR, Wireless, Cable, MPLS VPN and Metro Ethernet) in an enterprise environment.

• To implement security features across WAN setups for a secure communication.

1.5 Structure of the Report

The report consists of the following major parts:

Chapter

• Literature Review.

• Network Design.

• Results and topology tests.

• Discussion.

• Conclusion/Further work.

• References.

• Appendix.

CHAPTER NO.2

LITERATURE REVIEW

2.1 VoIP Fundamentals

2.1.1 A Brief History of VoIP

VoIP is still a relatively new technology. Enthusiasts began to recognize the potential for VoIP in 1995, when they found ways to avoid long distance charges by sending voice data packets over the Internet. Later that year, the first IP phone was created with the release of Internet phone software that could run on a home PC with sound cards, speakers, microphone, and modem. The software made acceptable quality voice calls if both users had the same software. The initial VoIP conversations had poor sound quality and connectivity but were still promising. Entrepreneurs soon realized that IP-based telephony networks and infrastructures could replace the more expensive PSTN networks at a fraction of the cost. By 1998, some investors set up a way to send voice calls by PC-to-phone and then by phone-to-phone. The introduction of broadband and Ethernet services provided greater call clarity and reduced latency. Although there were still problems associated with call static and dropped connections between the Internet and the PSTN, startup VoIP companies began to offer free calling service to customers from special locations.

In 1998, a handful of companies began to produce switching equipment, which resulted in a slight increase in VoIP usage. The VoIP switching equipment provided another device that could control the functions previously maintained by a PC's Central Processing Unit (CPU), including switching a voice packet into data the PSTN reads (and vice versa). This major enhancement made VoIP hardware less computer-dependent. When VoIP hardware became more affordable, larger companies implemented VoIP on their internal IP networks, and long distance providers routed some calls on their networks over the Internet.

2.1.2 Legacy Voice and Data Network

The term legacy refers to those networks that have two split physical design structure for voice and data network services. So companies's most of IT budget goes towards maintaining the distinct voice and data networks, which requires human resources for the maintenance of both the data and voice infrastructures. There are two other drawbacks associated with this approach is that these two separate infrastructures cannot share their resources with each other and have to be managed separately, both of which increase IT budget.

2.1.3 Overview of VoIP Technology Operation (Packetized Voice) and Future Multiservice Networks

Since when the advent of new technologies the demand and desire of the industry is also changing. So the combination of distinct voice and data networks led to the development of several new concepts and technologies, such as packetized voice.Packetized voice comprises several standards and protocols. Applications use these protocols and standards to provide value-added and cost-effective services to users. Wendell, Michael J. & Cavanaugh (2005).

Packetized voice enables a device to send voice traffic (for example, telephone and fax) over an IP/Frame Relay/ATM network. In case of (VoIP), the digital signal processor (DSP) that is located on the voice gateway segments the voice signal into frames. The voice gateway combines these frames to form an IP packet and send the packet over the IP network. On the receiving end, a reverse action converts the voice information that is stored in the IP packet into the original voice signal.

Across the IP network these voice packets are transported by suing the Real-Time Transport Protocol(RTP) and RTP control protocol stack and by using the User datagram Protocol(UDP) as a transport protocol.RTP provides timestamps and sequence numbers in each packet to help synchronize the voice frames at the receiving side.RTCP provides a feedback mechanism that informs session participants of the received quality of the voice call and includes information such as delay, jitter and packet loss. Sandy, G.. (2007)

It is important to note that most of the real-time application use UDP as the transport protocol rather than TCP, for the following reasons:

1. TCP guarantees the retransmission of frames that are lost in the network, which is of n use in a packetized voice network because the late arrivals of frames at the receiving end introduce delay.

2. TCP introduces unnecessary delay by waiting for acknowledgements for every packet. This delay is not noticeable in the data networks but cause poor quality in packetized voice networks. Kevin Wallace (2005).

In addition to using RTP/UDP/IP as the protocol stack to carry voice calls across the IP network, VoIP networks use VoIP signaling protocols to set up and tear down the calls which will be discussed in coming section. Introducing "packetized voice" capability into the router shown in figure-3 turns the router into a voice gateway and enables the router to do the packetization just described along with the duties required as a regular data router. This change allows providing additional services such as toll bypass.

Toll bypass enables to reduce the overall telephony expenditure by routing long distance interoffice calls over existing packet-based WANs, thus avoiding interexchange carrier (IXC) toll charges. To ensure that the design of toll-bypass application is effective, it must include a voice gateway that is able to support multiple types of call signaling as it interacts with the PSTN, WAN, and existing PBX systems.

Codec is also called Vcoder because it performs voice conversion from analogue to digital and digital to analogue signals. Numerous codec schemes are used in IP telephony such as G.711, G.723.1, G.726, G.729 etc. These schemes are based on different standards and algorithms, therefore yields different results. Because of diversity, each codec scheme has different level of speech quality, network bandwidth utilization and computation complexity.

2.1.4 Call Control in Circuit and Packet-Switched Networks

In traditional voice networks, each call consumes a fixed amount of bandwidth. The PBX does not place more calls than it can handle through the trunks connecting to the PSTN, as shown on the left side of figure-5. In packetized networks, if bandwidth is available to make only two good-quality calls, in the absence of a call admission control (CAC) mechanism, the voice gateway allows the third call to go through, as shown on the right side of figure-5. This third call degrades the quality of the existing two good voice calls. Hence, gatekeepers are deployed in the packetized networks to control the number of calls that can be sent over the WAN links. Wendell, Michael J. & Cavanaugh (2005).Gatekeepers perform CAC and bandwidth management in VoIP networks. Gatekeepers ensure that enough bandwidth is available before granting permission to a gateway to place a call across the IP WAN. Michael J. & Cavanaugh (2005) further discusses that after receiving permission from the gatekeeper, the originating gateway initiates a call setup with the terminating gateway over the packet network.

Besides performing CAC and bandwidth management, gatekeepers can perform accounting, call authorization, authentication (via RADIUS), address lookup and resolution, and translation between E.164 numbers and the IP addresses. Michael J. & Cavanaugh (2005).

2.1.5 VoIP Protocols

To understand VoIP, one must understand the protocols that drive this technology. This section of the report provides a brief overview of the operation of VoIP protocols and explains their functionality. There are two types of VoIP protocols:

• Protocols that carry voice payload (RTP, RTCP, UDP and IP)

• Protocols that provide the call control and signaling.

2.1.5.1 Internet Protocol (IP), UDP and Packet Transmission Mechanism

IP, or Internet Protocol, is fundamental to transmitting voice over the Internet. IP is a routing protocol that specifies packet formats, their addressing scheme, and the IP header, which contains information about the data transmission. The main purpose of IP is to route data from the originating to the terminating location (this function is also the primary purpose of a router). The term "layer" refers to the Open Systems Interconnection (OSI) model, a networking structure for implementing protocols in seven layers where control passes from layer to layer. Figure-5 shows how the OSI model correlates with the IP suite. IP correlates with Layer 3 of the OSI model. IP works with higher-layer protocols for systems to function correctly. Also, IP does not have any physical transmission equipment, so it works together with lower-layer transmission systems. In the OSI Model and IP Suite (see Figure-5), application data passes to the Transport Layer, where the packet receives a Transmission Control Protocol (TCP) or User Datagram Protocol (UDP) header. Wallace, K. (2005).The packet passes to the Network Layer, where the packet receives an IP header. The packet then passes to the Data Link Layer, where it becomes packaged for transmission. At the termination, each layer extracts its own header data and passes the remaining data to the above layer until the terminating system receives only the original data.

UDP on the other hand is a connectionless protocol run on the top of IP. It does not provide guaranteed data transmission therefore it is not suitable for Data. It has a minimal operational over head since it don't share messages like TCP do, therefore UDP is used to carry actual voice traffic in VoIP. UDP delivers the packet regardless of the packet loss, as voice is more sensitive to latency than packet loss. Wallace,K.(2005)

2.1.5.2 RTP and RTCP

The dominant signaling protocol used in IP-based networks is the Real-time Transport Protocol (RTP).It is defined by the IETF in RFC 1889.RTP is the Internet-standard protocol for transporting real-time data. RTP contains a data part and a control part. The control part is RTCP. RTP provides real-time point-to-point delivery of data, including audio and video. Systems usually run RTP on top of UDP to use UDP's multiplexing and checksum features. RTP does not provide a method for ensuring timely packet delivery or provide a guaranteed quality of service. Sandy, G.. (2007). Instead, it depends on the lower OSI layers to provide this function. RTP does not guarantee delivery or prevent packets from arriving out of order. It also does not assume a system is reliable and therefore delivers packets in sequence. Sandy, G.. (2007) further said RTCP is based on periodically sending control packets to participants in a session, using the same distribution techniques as the data packets. RTCP supports real-time conferencing within the Internet, including source identification and gateway support for audio and video bridges. Systems use RTCP to monitor the quality of RTP sessions, such as packet counts, packet loss, delays, and interarrival jitter. RTCP has a separate flow from RTP using the RTP port number + 1 for the UDP transport. RTP ports are always even; RTCP ports are always odd.As explained above the voice is suitable for UDP/IP due to its time sensitive nature but it need more information than what UDP contain therefore it uses RTP where IP packet is encapsulated in UDP and UDP inside the RTP packet header. The IP/UDP/RTP is 40 bytes header as show in the following figure.

2.1.5.3 Call Control and Signaling Protocols

The job of the call-control and signaling protocols is to setup and tear down the connection between two or more endpoints in a VoIP network. There are five main protocols for generating traffic: Session Initiation Protocol (SIP), Media Gateway Control Protocol (MGCP), Megaco/H.248, H.323 and Skinny Client Control Protocol (SCCP).

A). SIP

SIP is an RFC standard (RFC 3261) from the Internet Engineering Task Force (IETF).SIP is one of the most talked about protocols these days because of its wide interoperability with other protocols. SIP is a signaling protocol used to establish sessions in an IP network. A session can be a two-way call or a multimedia conference. SIP works with other protocols to describe session characteristics to potential users. SIP could use any type of transport protocol for media, but it typically uses RTP. SIP signaling and RTP media pass across a network separately from each other. SIP is a client-server protocol. SIP endpoints are capable of working as client as well as server where endpoints are called UAC (User Agent Client) and UAS (User Agent Server). SIP client includes phone gateways and server includes Proxy Server, Redirect Server, User Agent Server and Registrar Server they are explained below:

Proxy Server - a server that resides between a client application and a real server. It intercepts requests to the real server to determine if it can fulfill the requests. If it can, the proxy server completes the requests. If it cannot, it sends the request to the real server.

Redirect Server - a server that receives a request and provides information back to the user about the party a user wants to contact. The user can then communicate directly with the contact after receiving the information from the redirect server. The redirect server is no longer involved as opposed to the proxy server where two parties can communicate through the server.

User Agent Server - a server that contacts a user when it receives a SIP request and returns a response on behalf of the user. In reality, a SIP device functions as a user agent client and user agent server. The user agent client initiates a SIP request. The user agent server receives and responds to the SIP request.

Registrar - A user agent sends a registration message to the SIP Registrar, which stores the registration information and then sends a response back to the user agent. SIP is a text-based protocol that uses requests and responses to communicate within a network and establish connection between end points. A unique SIP address identifies a SIP user in a network. The SIP address is similar in structure to an e-mail address:

sip:jane.doe@company.com

Users register through a registrar server using their SIP address. The registrar server in turn provides the information to a server when requested. When a user makes a call, a SIP request travels to the SIP server through a proxy or redirect server. The request includes the address of the caller and address of the calling party.

SIP supports five facets of establishing and terminating multimedia communications:

User location - determines the end system to use for communication

User availability - determines the willingness of the called party to engage in communications

User capabilities - determines which media and media parameters to use .

Session setup - "ringing," or establishing session parameters at both the called and calling party

Session management - includes transfer and termination of sessions, modifying session parameters, and invoking services.

B). MGCP

It is standardized by the IETF in RFC 3435 and is based on a client/server model. As defined in RFC 2705,"MGCP is designed as an internal protocol within a distributed system that appears to the outside as a single VoIP gateway." MGCP is a protocol handled by a Media Gateway Controller (MGC) and a Media Gateway (MG). MGCs, also called call agents, control the operation of the MGs. According to Foster (2003) "MGCP is an application programming interface and text-based master/slave protocol used by the media gateway controller (MGC) or Call Agent (CA) to control media gateways (MG)". MGs have endpoints where the MGCs can create, modify, and delete connections to establish and control media through other multimedia endpoints. GCs handle call control and call signaling, and their commands usually relate to establishing connections and tearing down connections from one side of the MG to another. For example, an MGC uses MGCP to tell the MG to create a connection from a line or trunk on the circuit-switched side of the MG to an RTP port on the IP side of the gateway. Note that two MGCs can also communicate with each other with another protocol such as SIP or H.323.

C).Cisco Skinny Client Control Protocol (SCCP)

Cisco phones typically use the Skinny Client Control Protocol (SCCP) for call setup. SCCP, called Skinny, provides a simple, lightweight call setup protocol for endpoints controlled by Cisco Unified Call Manager (now being called Cisco Unified Communications Manager). Skinny passes messages using TCP and port 2000.

D). H.323

H.323 is an International Telecommunication Union (ITU) specification to carry real-time voice traffic, such as telephone calls, video and data over an IP network. It was basically made for LAN multimedia conferencing, but later was extended to VoIP. It does not guarantee the quality of service (QoS). It deals with call signalling, call control, point to point and point to multi point conferencing and also bandwidth utilization control. H.323 operation can be defined using four components including Terminals, Gateways, MCUs (Gatekeeper and Multipoint Control Units).

Terminal - an endpoint for real-time communication with other H.323 endpoints, usually an end-user communication device.

Gateway - an endpoint providing translation between an H.323 network and another type of network. One side of the gateway supports H.323 signaling and terminates media packets. The other side of the gateway interfaces with a traditional circuit-switched network and supports transmitting and signaling protocols used in the circuit-switched network. Translation occurs internally in the gateway.

MCU - provides conference support for three or more H.323 terminals. All terminals in the conference have a connection with the MCU. MCUs handle conference resources, work between terminals to determine audio or video coder/decoders (Codecs), and handle media streams. Another H.323 component is a gatekeeper. It is optional within an H.323 network. The gatekeeper allows network access from one or more endpoints permitting or denying a call from an endpoint within its control. Gatekeepers, gateways, and MCUs are separate components but can function as a single physical device.

The signaling messages exchanged between H.323 elements are part of H.225.0 and H.245:

H.225.0 Call Signaling -establishes and tears down connections between H.323 endpoints.

H.225.0 Registration, Admission, and Status (RAS) - used between endpoints and gatekeepers, enabling a gatekeeper to oversee the endpoints within its zone.

H.245 Control Signaling - control protocol used between two or more endpoints. Its main function is managing media streams between H.323 and session participants.

2.1.6 Hardware Equipments & Endpoints (Phones, Gateways etc)

There are number of devices that involve in the communication process, these include:

• Cisco IP Phones

• Soft phones

• Wireless Phones

• Voice Gateways

• Survivable Remote Site Telephony

• Call Manager Express

All these have specific features that are much useful for the communication.

2.2 VoIP Benefits and Problems.

The following section will highlight the benefits of VoIP for an enterprise and the challenges faced by VoIP. Further it will portray the need of quality of service for VoIP.

2.2.1 Benefits of VoIP for Enterprise

There are many benefits of VoIP implementation for the enterprise. One of the reasons why enterprises are migrating to the VoIP infrastructure is Cost. Cost savings remains the greatest advantage of VoIP because IP networks use bandwidth more efficiently — users only pay for the data they send. With traditional circuit-switched networks, the network must reserve a channel for a conversation even when no one is talking. Although early VoIP call quality was inconsistent, voice quality has increased over time and often matches the public voice network. VoIP is changing the way people communicate by combining voice and data into a single integrated platform based on packet technology. A part from the Cost saving this technology provides services required for voice communication and provides the following benefits for the enterprise and end users:

• Avoids toll charges by regular telephone services.

• Lowers equipment costs.

• Both voice and data functionality is integrated.

• Lowers bandwidth requirements.

• Increases revenues and profits with new applications, such as video calling, unified messaging, and web-enabled multimedia call centers.

• Increases mobility as users take advantage of wireless IP networks.

• More features—voice mail, caller ID, call conferencing, and call forwarding.

• Unlimited long distance.

• Integration with PC software.

2.2.2 VoIP Unknowns

The following summarizes some pending issues associated with VoIP:

• Hardware capability and configuration from lack of standard equipment.

• Security—regulatory agencies undecided over how to handle issues such as security breaches.

• Taxes—Although VoIP services are currently not taxed the way traditional phone. services are taxed, state and federal governments could levy universal service fees in the near future.

• E-911—VoIP providers must currently notify their customers about limited E-911 capability because some plans do not support this service.

2.2.3 Problems with VoIP and Why Quality of Service (QoS) is essential?

Although there are so many advantages of VoIP and due to these advantages enterprises are rapidly moving towards implementing this technology in there infrastructures but in spite of these advantages, there are so many challenges for VoIP as well. The biggest challenge to adopting VoIP services is to provide "service quality" and "reliability" equal to existing PSTN circuit-switched networks. As explained in the first section of the report that IP faces a different challenge than PSTN, such as delivering voice packets in sequential order. When using VoIP, systems transmit IP packets in any order. The receiving VoIP system must reorganize the packets in sequential order and verify no packets are missing i.e. no packet loss. This difference in the way voice packets transmit means VoIP also has to verify and deliver a certain level of Quality of Service (QoS), as well as avoid latency, jitter that can affect VoIP calls differently than calls placed over the PSTN.

Quality of service (QoS) implementation is essential for today's converged networks, where data, voice and video traverse the network using same path Sandy, G.. (2007). It is really a challenging task to handle the application with different kind of sensitivities e.g. voice traffic has sensitivity to end-to-end delay and an acceptable delay could be between 150 to 200 milliseconds on the other hand an FTP file download has no such delay requirements. The problem arise in the networks with limited bandwidth and stringent requirements, it normally happened with WAN links where propagation of end user data is the first priority. Tim, S(2006 said to overcome the problem of limited bandwidth there must be some sort of control mechanism, which ensures the availability of bandwidth to delay sensitive traffic like voice and video. QoS is the mechanism that can ensure the prioritized traversal of sensitive traffic; the basic purpose of the QoS is to solve traffic performance issues.

2.3 Implementing VoIP Support in an Enterprise Network and Deployment Models

This section of report will give an overview of telephony deployment models, their necessary elements and components in an enterprise network. It also discusses how to implement voice in enterprise networks in detail by taking the example of an enterprise with three branch offices namely A, B and C. Further it briefly introduces Cisco CallManager and its different implementation options. Although Cisco Call Manager is not used during the implementation of this project, only its condensed version Cisco Call Manager Express is used which contains the basic functionality, features of Call Manager.

2.3.1 Enterprise Voice Implementations

The main telephony elements of an enterprise VoIP implementation are gateway, gatekeeper, Cisco Unified CallManager, and Cisco IP phones. Cisco IP phones need CallManager, because it acts as an IP PBX for the Cisco IP phones. The gateways provide connectivity between analog, digital, and IP-based telephony devices and circuits. Gatekeeper is an H.323 device that provides call routing or CAC services.

Enterprise voice implementations can vary based on many factors. One of those factors is the

number of sites, and the preferred method of data and voice connectivity (primary and backup)

between the sites. Some sites might have VoIP connectivity but no IP phones or other IP Telephony services.

At Branch A, IP Telephony services and IP phones have been deployed. Branch A has a Cisco

Unified CallManager cluster, and all employees use IP phones. Branch A is connected to Branch

B using a metropolitan-area network (MAN) connection such as Metro Ethernet; voice calls

between Branch A and Branch B must use this path. The Branch A connection to Branch C is over a WAN, such as legacy Frame Relay or ATM (a modern connection would be an MPLS VPN connection); voice calls between Branch A and Branch C must use this path. If WAN or MAN connections are down, voice calls must be rerouted via PSTN; if there is congestion, using the automated alternate routing (AAR) feature, voice calls are again rerouted via PSTN. Wendell, Michael J. & Cavanaugh(2005) .Note that at Branch A, voice calls to and from people outside the enterprise are naturally through PSTN At Branch C, on the other hand, the old PBX system and phones are still in use. Wendell, Michael J. & Cavanaugh(2005) further said that if the MAN connection goes down, survivable remote site telephony (SRST) deployed on the Branch B gateway allows Branch B IP phones to call each other, but calls to anywhere else are limited to one at a time and are sent over PSTN.

2.3.2 Enterprise IP Telephony Deployment Models

By using CallManager, organizations can eliminate PBX and replace it with IPT over a converged network. CallManager provides call-control functionality and, when used in conjunction with IP phone sets or a soft phone application, it can provide PBX functionality in a distributed and scalable fashion. Wallace,K.(2005). The option that is suitable for an enterprise depends on the organization of that enterprise, its business strategy, budget, and objectives. The four main options are as follows:

1. Single site

2. Multisite with centralized call processing

3. Multisite with distributed call processing

4. Clustering over WAN

A). Single-Site Model

In this deployment model, shown in Figure -15, CallManager applications such as voice mail, IP-IVR, AutoAttendant (AA), Transcoding, and conferencing resources are located at the same physical location. All the IP phones are located within this single site. The PSTN is used to route the calls going outside.

B). Multisite with Centralized Call Processing Model

Figure-16 shows the centralized call-processing deployment with remote branches. All the call processing is done at the central site. This is suitable for organizations in which the majority of the workforce is concentrated at a single site and small numbers of employees work at the remote branches. At each remote branch, SRST routers ensure that call processing is preserved in case of WAN link failure. Wendell, Michael J. & Cavanaugh(2005).The voice traffic travels via the IP WAN and falls back to the PSTN if not enough bandwidth is available across the WAN link, by using the Automated Alternate Routing (AAR) feature available in the CallManager.

Wendell, Michael J. & Cavanaugh(2005) further said to avoid the oversubscription of the WAN bandwidth by using the locations-based CAC feature in the CallManager. This feature works for hub-and-spoke topologies.This deployment model is cost effective and provides many benefits, such as a unified dial plan, less administrative overhead, and potential savings on communications costs as intersite calls use the IP WAN as first choice. The only limitation is that the remote sites will have limited features available if a WAN failure occurs when operating in the SRST mode.

C).Multisite with Distributed Call Processing Model

In a distributed call-processing deployment, CallManager and applications are located at each site. Figure-17 depicts a distributed call-processing model in which headquarters and branch A IP phones are served by separate CallManager clusters and branch B is served by the Cisco CallManager Express (CME) feature that is enabled on the router. CME solution is suitable for a small branch. Intercluster Trunks link CallManager clusters. Wendell, Michael J. & Cavanaugh(2005) further said the Cisco IOS gatekeeper ensures that only a permitted number of calls are sent across the IP WAN between the CallManager clusters. PSTN connections on the gateways route the local off-net calls from each location and serve as a backup connection to IP WAN when insufficient bandwidth is available to support additional calls.

D).Clustering over WAN Model

The vital activities in any business are continuity planning and disaster recovery. Large and small disasters happen all the time. The Cisco IPT solution allows organizations to build disaster recovery sites by separating the single CallManager cluster across the WAN. CallManager servers in a cluster update the configuration information via the Microsoft SQL replication process. To ensure successful SQL replication and propagation of other critical information in real time, the round-trip time (RTT) between any CallManager servers in the cluster should not exceed 40 ms.

When using the clustering over the IP WAN deployment model, deploy voice gateways, media resources, and voice mail locally at each site. Essential services such as DHCP, DNS, and TFTP that are critical for the functioning of IP phones and other IPT endpoints must also be deployed locally. This configuration avoids dependency on a single site for crucial resources.

2.4 Quality of Service (QoS)

In networking, QoS describes a large array of concepts and tools that can be used to affect the packet's access to some service. Most of us think of queuing features when we think of QoS, reordering the output queue so that one packet gets better service than another. But many other QoS features affect the quality such as compression, drop policy, shaping, policing and signaling etc. Sandy, G.. (2007) define QoS as "the ability of the network to provide better or "special" service to a set of users/ applications to the detriment of other users/application".

QoS implementation is essential for today's converged networks, where data, voice and video traverse the network using same path. It is really a challenging task to handle the application with different kind of sensitivities e.g. voice traffic has sensitivity of end-to-end delay and an acceptable delay could be between 150 to 200 milliseconds.Ranjbar, A. (2008) said "QoS is the mechanism that can ensure the prioritized traversal of sensitive traffic; the basic purpose of the QoS is to solve traffic performance issues."

2.4.1 Important Characteristics of Traffic

To solve the traffic performance issues QoS uses some of the characteristics of the traffic these are listed below.

• Bandwidth

• End-to-End delay

• Jitter

• Packet Loss

2.4.1.1 Bandwidth

The term bandwidth refers to the number of bits per second that can reasonably be expected to be successfully delivered across some medium. In some cases, bandwidth equals the physical link speed, or the clock rate, of the interface. In other cases, bandwidth is smaller than the actual speed of link.

In PSTN bandwidth is guaranteed to each voice stream therefore there is no variable delay. However this is not the case with data network where bandwidth is shared among the connection and must be calculated to make it possible to have less delay, jitter and packet loss. Wallace, K. (2005).As PSTN provides 99.999 percent of network uptime in order to gain same level of quality of service in data network (IP), aspect such as delay, jitter and packet loss must be resolved and minimized.

2 Delay:

Delay or latency is the time when a voice packet is transmitted over network and the time it reaches to the destination. The delay can be measured by two methods including one-way delay and round-trip delay where one-way delay is the time a packet takes from source to destination it is one directional while round-trip delay is the time a packet takes to its destination and back to its source in the form of acknowledgment. One way delay require more resource while round-trip delay can be measured easily, if round-trip delay is divided by 2, one way delay can be achieved.

Scott, E. & Roth,H(2008) discusses about the type and said that there are mainly two types of delay, fixed delay and variable Delay. Where Fixed Delay as name emphasizes is fixed, predictable which add to the overall delay on the connection. There is several fixed delay such as coding delay which depends on the codec, involved in translation of audio signals to digital for transmission. Once the audio is converted to digital it is placed into the packet, caused packetization delay, is the time taken placing the voice into the packet and taking it out. Packet is to be placed on the link for propagation which cause serialization and propagation delay. Variable Delays on the other hand take place due to queuing delays and network delay. Following picture better represents these delays.

3 Jitter:

Jitter is the variations in expected arrival time of a packet at the receiving end of network it also known as delay variation. The more variation in the delay the more jitters in the network. Packet take different route in the network to the destination which create varying gap between the packets creating bad network performance with respect to voice as they played out at their arrival. There is mechanism to overcome this shortcoming, namely jitter buffer, which is used at the receiving end to store the packet and play them out in a sequence producing good quality of voice. The packet at receiving end placed in centre position of the jitter buffer in a manner that the last packet is placed behind the first packet. It poses some concern such as if a packet arrives early than its expected time the buffer will drop the packet and if it is delayed, the buffer will simulate some kind of sound or allow period of silence. The packets are moved from centre to the end of the buffer for play out causing another type of delay, namely jitter buffer delay as showed in the figure-19. Therefore it is imperative to know which kind of jitter a network is being suffered from in order to provide quality of service. Jitter is measured by the receiving device, each RTP packet header contains time stamp and the receiver system clock are used to compute delay variation which is reported by the RTCP and sent to the transmitting device in every five seconds.

4 Packet Loss:

Packet loss is a rate of losing packets along the way to the destination. It could happen either in the network or at the receiving end. It occurs when a network device such as router has no more space to hold the incoming packets and it ends up dropping them. A lost packet can be recoverable through retransmission but due to the nature of voice traffic which should be played out in real time, is inappropriate as shown in the figure.

Sandy, G. (2007) said that packet loss degrades the voice quality by introducing gaps in the conversation which cause voice clipping. There is process which is used to fill the gap by either sending lost packet again or create estimated packet to improve the quality, degraded by the lost of packets. This is called packet loss concealment (PLC).

Following few lines determine network performance requirements for voice and video traffic Tim, S(2006):

• Voice:

  1. No more then (150) ms of one-way delay.

  2. No more then (30) ms of jitter.

  3. No more than 1 percent packet loss.

• Video:

  1. No more than (150) ms of one-way for interactive video application (video conferencing).

  2. No more than (30) ms of jitter.

  3. No more than 1 percent packet loss.

2.4.2 QoS Implementation Methods:

There are different implementation methods used commonly, they are explained briefly as under:

2.4.2.1 Legacy Command-Line Interface (CLI):It is most complex and time consuming method. It requires learning different syntax for each QoS mechanism. Because of its manual nature it is more error prone and time consuming.

2.4.2.2 Modular QoS Command-Line Interface (MQC): It is a modular command-line interface that is common across different Cisco platforms and it separate the task of defining different traffic classes from the task of defining QoS policies.

There are three major steps while implementing MQC the steps are defined below.

• Define traffic classes for different type of traffic; this will divide the network traffic into some specific classes.

• Define QoS policy for different traffic classes; policy defines the treatment that should be done at the router when the traffic starts passing the router.

• The third step is to apply that particular policy to an interface, sub interface or circuit. The policy could be applied in any direction.

2.4.2.3 AutoQoS:It automatically generates QoS commands, it is simplest and fastest method. But it has limited ability to fine-tune results. One has to fine-tune its results.

2.4.3 Classification and Marking:

Classification and marking is the core for providing QoS to different types of traffics. Different traffic types are classified into different service classes in order to provide them some sort of treatment based on the class that traffic belongs. Ranjbar, A. (2008). The classification and marking tools not only classify different traffic into classes but it also mark some fields in the header of traffic for some value so that other QoS tools can classify the packet easily after examining the marked fields in header.

2.4.3.1 Classification with NBAR:

Network Based Application Recognition is service that can help classify traffic into different classes; it is used to classify the traffic that is difficult to classify Ranjbar, A. (2008). NBAR recognize the application specific information by looking into the payload beyond the TCP or UDP header, NBAR also allow searching for Citrix application types and a portion of URL string. Even if NBAR is not used for classification of traffic it can be used to gather statistics about packets entering and exiting the interface Ranjbar, A. (2008).

2.4.3.2 Marking:

Marking is the process of setting some values inside the data link layer header or network layer header for the purpose of recognition by other device's QoS tools. Tim, S(2006) said marking is to reduce the overhead of defining classes of traffic on every single device throughout the path rather it marks the field inside the header with some specific value and every device in the way treat the traffic accordingly.

2.4.3.3 Class-Based Marking:

Class-Based Marking can also be used for the classification of traffic into different classes; CB Marking can classify the traffic by directly examining the header of the packet, frame etc. CB Marking can also use access lists for the classification of traffic, and IP access lists can match source IP address, destination IP address, UDP port number, TCP port number and many more for the classification of traffic. CB Marking can work without ACLs and can match many different criteria like match protocol, source MAC address, CoS, IP precedence, DSCP value in the packet etc Ranjbar, A. (2008).

2.4.3.4 Layer 2 Marking:

Different Layer 2 protocol data units (PDUs) have different fields for QoS marking. On 802.1Q/P or ISL frames, (3) bit PRI Class of Service (CoS) field is used. Frame Relay has backward explicit congestion notification (BECN), forward explicit congestion notification (FECN) and discard eligibility (DE) for congestion notification and discard eligibility notification. Scott, E. & Roth, H (2008)

2.4.3.5 Layer 3 Marking:

IP header or layer 3 headers does not change and remain intact throughout the network path, that's why it is more preferred to use layer 3 marking as compared to layer 2 marking. Layer 3 has two very commonly used marking schemes called IP precedence and IP DSCP, the first scheme uses 3-bit header field for marking of traffic while the other use 6-bit header field in IP header.

2.4.4 Congestion Management and Queuing:

In real networks it is very common to have congestion on a link, the reason could be any thing from speed mismatch (means ingress or input interface is of high speed as compared to the output interface) to aggregation problem or confluence problem, which means that multiple traffic streams joins here to make a big output stream.

To overcome the problem of congestion queuing technique is used in which the packets entering the router wait for their turn to get out from an interface, the time each packet have to wait depend on the queue type and priority of the packet. Some other queuing techniques include FIFO, Priority Queue, Custom Queuing, and Weighted Fair Queuing.

2.4.4.1 FIFO, PQ, RR and WRR:

Different queuing techniques have been developed and used for congestion management. FIFO is the default queuing mechanism for most of the interfaces. FIFO is a simple algorithm, which need no special configuration but class, priority and type of traffic does not have any effect on that queue Ranjbar, A. (2008).This type of queue does not work well with the delay sensitive traffic like video and voice because all the packets are treated equally without any differentiation.

Priority queuing (PQ) is solution for delay sensitive traffic; it has been used for many years and consists of four queues high, medium, normal and low priority queues. The configuration of the PQ is requirement and network engineer has to tell where to put the packets otherwise the packets are placed in normal priority queue. The scheduler takes the packets from the high priority queue first; when the high priority queue is empty one packet is taken from the medium priority queue and after one packet the scheduler move back to high priority queue if there is any packet waiting in this queue. If high and medium priority queues both are empty the scheduler take one packet form normal queues, in this way after every single packet the scheduler move back to high priority queue Ranjbar, A. (2008). PQ has this phenomenon of scheduling and serving the queues, the queue with the lowest priority is served only if all the high priority queues are empty, this cause PQ danger for starving lower priority queues. . The figure illustrates the concept of priority queuing.

Round Robin queuing is another solution, in RR few queues are created and packets are placed into queues. The scheduler takes one packet from the first queue second from the other and so on in this fashion RR queue serve the packets in every queue, the plus point of RR queuing is starvation free traffic movement and negative is that there is no prioritization of traffic .

Another modified version of the RR queue is weighted RR queue, in which every queue is provided some weight and that weight defines the allocated bandwidth share for the queue. Custom queuing is an example, where network engineer can set the number of bytes to be processed before moving towards the next queue.

WRR and CQ have their own weaknesses like if the bandwidth allocated to a queue is near or multiple of MTU then the queue sending the larger packets near to MTU will get the most of the bandwidth of the link.

2.4.4.2 WFQ, CBWFQ and LLQ:

Weighted Fair Queuing is another queuing mechanism very widely used for low bandwidth links; WFQ is a flow-based queuing algorithm. Arriving packets are classified into flows, and each flow is assigned to a FIFO queue. A flow can be identified by one of many fields in the header of payload like source IP address, Destination IP address, TCP/UDP port number or ToS field Ranjbar, A. (2008).. A hash value is calculated for every flow, and packet for the same flow end up with identical hash value and are placed in the same queue. Packets for new flows are placed in new queue for that flow. WFQ uses sequence numbering for scheduling purposes and priority and flow influence the sequence number of the packet. The figure given below will help understanding the operation of WFQ.

There are few benefits of WFQ, it is very easy to configure, it does not starve any of the flow, it also have some drawbacks like it cannot be configured or modified, it cannot guarantee bandwidth and delay to some specific traffic Sandy, G.. (2007).

Class Based Weighted Fair Queuing (CBWFQ) addresses some of the limitation of CQ, PQ and WFQ. Unlike WFQ it allows the creation of user defined classes and allocation of bandwidth guarantee defined by the user. It addresses the starvation problem of PQ and allows every queue to pass its packets, the user defined traffic classes are created by using class map statement, which is very easy to configure and user friendly as compared to access lists. CBWFQ guarantee the bandwidth availability but the main problem with this type of queue is that it can't handle delay sensitive traffic such as voice, so this kind of queue is not suitable for voice of IP (VOIP) traffic.

WFQ and CBWFQ can guarantee the bandwidth availability but cannot guarantee the low delay to specific applications like VoIP, applications like VoIP has very small tolerance to delay and jitter so that's why those application need quick response for service. WFQ and CBWFQ can't provide quick response because they don't have any priority queue Sandy, G.. (2007) . LLQ has a strict priority queue, which has a priority over other queues. It makes it very suitable for delay and jitter sensitive applications. The strict priority queue in LLQ is policed and is provided a low bandwidth, which means that other queue will not starve for bandwidth even if there is constant traffic in strict priority queue. LLQ is a CBWFQ with strict priority queue. The function of the low latency queue is shown in the figure below. LLQ is the only suitable queue for voice traffic because of its strict priority queue implementation, and is our obvious choice in our network design.

2.4.5 Congestion Avoidance :Policing and Shaping:

Traffic policing and shaping is used in situations where traffic of one network is entering into another network e.g. Internet service provider (ISP) network. For that particular kind of service there is an agreement between the user and service provider, which is called service level agreement (SLA) Ranjbar, A. (2008). The SLA defines the committed information rate (CIR) of the channel, traffic rate and billing matters. The bandwidth availability between the private network and ISP is limited because of expenses, therefore normally some sort of shaping is used at customer end and policing is used at the service provider's end Ranjbar, A. (2008)..

Shaping is somehow the process of artificially delaying the packets for better service, for example, if we have a link to a service provider with CIR of 64 Kbps and a link of 2 Mbps and no shaping is applied at our edge router. The router will try to send data at rate of 2 Mbps; the result will be extensive drop in our traffic because we are only guaranteed the delivery of 64 Kbps. Traffic shaping can solve this problem by buffering the extra traffic at edge router and sending the traffic in accordance to CIR.

The policing are applied at the inbound interface of the edge router of service provider, the policy could be based on CIR, and class based traffic and so on. The policy could be even to drop the excessive packets at an interface, or to remark them with another priority value and let the other networks to drop the packet Sandy, G. (2007). If the service level agreement is signed for class-based traffic policing the traffic shaping at the customer end should be class based to save the packets from dropping or remarking. Both shaping and policing are very strong features used to tune the QoS in networks to overcome the problem of congestion and delay Sandy, G.. (2007).

CHAPTER NO.3

NETWORK DESIGN

To start with this important bit of the report few assumptions should be made to understand the design concepts and strategy. This section of the report will illustrate the network design of a small to medium size enterprise that has two branch offices and one head office communicating through WAN clouds of Frame Relay and Wireless. Multisite with Distributed Call Processing Model is used with hub and spoke topology.

Due to non availability of Cisco Call Manager (CM) in LAB, Cisco Call Manger Express (CME) is used. Its a slimmed-down version of the Call Manager (CM) server application.CM runs on dedicated server, while CME runs on a router.CME possesses much of the basic functionality of CM, which may be all that is needed in a smaller network without a large number of phones.CME may also be much more cost effective in many environments where the full power of CM is not necessary.CM and CME both act as servers whose main function is to establish class between phones, as well as many other voice related functions

3.1 Enterprise Edge Design

This design is a simulated design of an enterprise in the LAB environment. Since the enterprise is small to medium size so a collapsed core block design is followed where the core layer is collapsed into the distribution layer. The distribution and core layer functions are performed by the same device. Collapsed core is not an independent building block but it is integrated into the distribution layer of the switch block.

The enterprise is connected to the external resources through its edge called the edge of the enterprise. These resources include internet access by one or more Internet Service Providers (ISPs), web servers and applications and connection to the WAN clouds of frame relay and wireless.

As shown in the following network diagram of the enterprise, one end of head office edge (HO-EDGE) is connected to the WAN cloud and the other end is connected to the internal network of enterprise. It is also connected to the internet cloud through the edge. Similarly on the other end across the WAN cloud branch office is connected to the WAN cloud through its branch office edge (BO-EDGE) and to its internal network.

Both Head Office Edge and Branch Office Edge internally are connected to Cisco Call Manager Express (CME). which performs all the call control functions.Thus enabling the voice communication through phones across the WAN connection. QoS mechanisms for the voice traffic are focused on the Service Level Agreement (SLA) between the service provider and enterprise. Some measured parameters like committed information rate (CIR), committed burst (Bc) and excess burst (Be) are been negotiated and ensured. All the policing are applied at the outbound interface (egress) of the edge routers of enterprise e.g. policy maps and class maps are implemented on these interfaces of head office and branch office.

3.2 Frame Relay Design

3.2.1 Benefits and Justification of the Design

Frame relay is a connection-oriented data link layer communication and itsPermanent Virtual Circuits (PVC) are 24 hours 7 days a week technology at very cheap cost. PVCs are the frequent and reliable data transfer between the two end devices across the Frame Relay cloud. Frame Relay is distance independent technology so there is no constraint for distance between the branch offices and the head office. PVC can be useful for all broadband data, voice and video traffic. Frame Relay doesn't put overhead on the network because it has congestion notification mechanism like Forward-Explicit Congestion Notification (FECN) and Backward-Explicit Congestion Notification (BECN). Committed Information Rate (CIR) proved to be reliable data transmission channel and Committed Burst (Bc) and Excess Burst (Be) also make data or voice efficient and reliable. The Service Level Agreement defines the committed information rate (CIR) of the channel, traffic rate and billing matters. As is this case it is fixed as 64kbps between the service provider and enterprise. The only factor that we can't ignore is, Frame relay is shared lines so insecurity can be a possibility, and when voice over this network will be implemented the security issues will be taken under complete consideration. Many security features can be used by other network layer security features like IPSEC or PPTP tunneling of VPN or other third party sources.

Distributive call processing design will make the long term future scalability to the network. Call control systems are available in head office as well as branch office. So there is no need to switch the calls through the head office as in the case of centralized model. This will not only reduce the time and network overhead as well as it is more reliable communication. Hub and spoke topology is used for data traffic as head office is only connected to outside world i.e. (internet) thus making the network more reliable. Secondly all the servers like web server, file server etc are in head office so there is minimum chance of data security threats and issues.

3.2.2 Implementation Steps

The following points will enlighten the implementation steps in detail as per the network design:

1. Build and connect routers, switches and PCs according to the network diagram. The frame relay routers should have DCE interfaces.

2. Assign the IP addressing to the interfaces according to the following table.

1. Configure routing protocol OSPF on all routers except Frame Relay (FR) routers.

2. Configure the FR1, FR2 and FR3 as frame relay switch.

3. Enable encapsulation as FR on the Head Office (HO) edge and Branch Office (BO) edge serial interfaces.

4. Attach DLCIs to the interfaces of Head Office (HO) edge and Branch Office (BO) edge.

5. Route DLCIs on the interfaces of FR1, FR2 and FR3 accordingly.

6. Configure the CIPC soft phones on the PCs attached to CME1 (Cisco Call Manager Express) and CME2.

7. Connect two traditional POTS analog phones on FXS0/0/0 ports of CME1 and CME2.

8. Check the connectivity of the network by ping command.

3.2.3 Implementation and verification of Quality of Service (QoS)

QoS structures are implemented on traffic.So class map and policy maps are designed for the QoS in order to discuss these, first we need to do the audit of the network by NBAR which detect and identify a wide range of protocols and applications. To run NBAR protocol discovery, CEF should be enabled on the (F0/0) interfaces of (CME1 and CME2) the following screen shot give details of the protocols currently running with the command:

The marking is done based on the scenario. First we need to mark the traffic with QoS markings criteria in which Voice Data is marked as dscp (expedite forwarding),Voice Control is given assured forwarding value (41) while FTP is given dscp-default assigned i.e. best effort.

To classify the traffic we configure class maps with certain conditions in this case it is implemented by matching the protocols as given below:

Three classes are defined and they are matched against certain conditions of the protocols. To check the class maps we can apply the following command on the edge routers:

To apply the policy we create policy map so it is created named (Project_Policy) and markings are assigned. To verify the policy map just created with the marking of the traffic by applying following command:

sh policy-map command on the edge routers i.e(HO-Edge and BO-Edge)

Now the last step is to apply the policy to the outgoing interface i.e. (s0/3/0) of HO-Edge and (s0/3/1) of BO-Edge with the service policy output command.

3.2.3.1 Configuration and Verification of CBWFQ:

In CBWFQ we will assign the bandwidth to certain class of traffic in this report two types of class traffics are assigned the bandwidth value they are given below:

Class Voice_Data

Bandwidth 30

Class FTP

Bandwidth 16

These two classes are nested in the policy map (Project_Policy) and then should be applied on the outer interfaces as explained above of (HO-Edge and BO_Edge), same configurations are applied on both the edge routers interfaces.We can verify this by the show command as below:

HO_edge>sh policy-map Project_Policy

3.2.3.2 Configuration and Verification of LLQ:

The only difference between LLQ and CBWFQ is LLQ adds a priority queue to the CBWFQ system. In our case Voice traffic is most important so that's why it is given priority and has been assigned a value of (50),instead of bandwidth command in CBWFQ we will put priority command that specify the a percentage of the total bandwidth. We can verify that with the sh policy-map commands on the edge routers e.g.:

3.3 Wireless Design

Wireless design is also a simulated design in the LAB environment and it is almost similar to that of Frame Relay.Linksys access points are used as a wireless medium between the head office and branch office. Access points are connected to the HO-Edge and the BO-Edge. As wireless is a shared media one cannot guarantee the bandwidth all the time so a Service Level Agreement(SLA) will be needed between the enterprise and the service provider to ensure the bandwidth and connection availability. Another important thing that should be part of the SLA is QoS implementation design for the voice traffic both on the service provider edge as well as on the enterprise edge. This will ensure the voice traffic to travel across the wireless without any delay as some priority is assigned to it. Wireless networks are always prone to network security issues and threats. So appropriate security features should be employed to overcome these issues and threats. It will make the communication reliable and more secure.

3.3.1 Implementation Steps

Implementation steps are almost same as that of the Frame Relay with some addition and subtraction. All the points related to FR are ignored and following points should be added as additional steps for wireless implementation.Linksys access points >WAP54G (EU/LA) 802.11gare used. They specify a maximum data transfer rate of 54Mbps, an operating frequency of 2.4GHz, and backward compatibility with 802.11bdevices. The Wireless-G Access Point has been designed for use with 802.11g and 802.11b products. The Access Point is compatible with 802.11g and 802.11b adapters, such as the Notebook Adapters for laptop computers, PCI Adapters for desktop PCs, and USB Adapters for USB connectivity. These wireless products can also communicate with a 802.11g or 802.11b Wireless PrintServer. To link a wired network with wireless network, connect the Access Point's Ethernet network port to any switch or router. Additional steps are described below:

1. Connect two linksys access points to the Fast Ethernet interfaces of HO-Edge and BO-Edge.

2. Assign the IP address to the access points and interfaces as under :

1. Enable the AP mode as bridge mode on both access points.

2. Swap the MAC address of APs with each other and enter them in remote bridge MAC address fields.

3. Ensure that SSID and Channel number on both sides are same to enable communication.

CHAPTER NO.4

RESULTS

This section of the report is dedicated to the illustration of all the results obtained during the LAB work in the implementation and testing phase of the project. Different tests are conducted by building different topologies. This section will also give the critical comparison of VoIP implemented over WAN technologies (FR and wireless) by making different test benches and altering (codec, bandwidth, web traffic, making two simultaneous calls etc) and most importantly implementing the QoS techniques like (CBWFQ and LLQ).Further it will analyze the results considering three important parameters i.e. packet loss, jitter and delay.

Different network analysis softwares are used for the analysis of the results e.g. VQ Manager and Wireshark.Both give different analysis options for different traffic patterns. Numerous captures are reserved and analyzed and will be discussed in detail in the following section.Xitami web server is used to generate web traffic across the network. It's fast, smart and robust and supports HTTP/1.0, FTP traffic.

Cisco switch feature SPAN is used to mirror traffic from one interface (source) to the another interface (destination), no other traffic is allowed on the destination interface port. SPAN is very useful tool for monitoring the state of systems on a network.

To start this important section first of all different show commands and call connection notification messages for (H.323 and Skinny) are important to explain and to ensure the call connectivity and establishment of analog and soft phones across WAN links.

Branch_Office1#sh voice call summary

PORT CODEC VAD VTSP STATE VPM STATE

0/0/0 - - - FXSLS_ONHOOK

0/0/1 - - - FXSLS_ONHOOK

0/2/0 - - - FXOLS_ONHOOK

0/2/1 - - - FXOLS_ONHOOK

50/0/1 .1 g729r8 y S_CONNECT EFXS_CONNECT

Branch_Office#sh dial-peer voice summary

dial-peer hunt 0

AD PRE PASS OUT

TAG TYPE MIN OPER PREFIX DEST-PATTERN FER THRU SESS-TARGET STATPORT

20001 pots up up 2222$ 0 50/0/1

1111 voip up up 1111 0 syst ipv4:192.168.20.1

8888 pots up up 8888 0 up 0/0/0

4444 voip up up 4444 0 syst ipv4:192.168.20.1

In the below screen shoot H.225.0 Call Signaling - establishes a call between H.323 endpoints. It is explained with the red arrow marks. The messages like setup open logical channel is sent and it is acknowledged then call proceeding channel is opened then it is acknowledged and then connect message, connect the two end point and last message is sending notify message to both ends and then voice stream starts.

Similarly SCCP (Skinny) protocol also established connection between two end points with some notification messages as explained in the screen shoot. First it open the channel then prompt status message then set lamp message then on the other side these message are acknowledged then call information is shared and tone is stopped and media stream starts.

4.1 Test Topologies for Frame Relay.

Different test topologies were build and tests were performed in LAB and are explained below:

4.1.1 Test Topology-1 (Simple Network)

Following table will illustrate the IP addressing:

Conditions/Scenario for this test:

• One important condition for this test is that Bandwidth is fixed as 128Kbps between the two CMEs.

• Wireshark PC is placed in the branch office and web server is in head office.

• VQ Manager is also placed for live call captures.

The following VQ Manager Screenshot will show the voice traffic across the network along with other traffic

There is another VQ Manager Screenshot which is kept during a progressive call. It is showing one LIVE call as well as jitter, packet loss and MOS (Mean Opinion Score).The main advantage of using the VQ manager is it gives the live call captures after every 10 sec or what ever time you will set and it keep you informed through its graphical charts about the current status of various voice quality parameters. From the capture below one can see the changing states and values of jitter, packet loss and MOS.

Numerous wireshark captures have been reserved while making a call through softphone from head office to branch office, with the flow of web traffic from head office to branch office.

One of capture's result is very very interesting and its graph is given below, it gives the jitter value almost (330ms) with heavy web traffic across the network. The reason for this peak value of jitter is heavy burst of web traffic flowing across the network before making the call. So when the call is made the voice traffic buffers (queued) behind the web traffic on router as there is no priority or quality of service applied to the voice traffic and by default WFQ is operational on router and doesn't guarantee the bandwidth and delay. This value of jitter in any case is not acceptable as the acceptable value for jitter is (30ms).

During the analysis RTP packet loss statistic observed is also not acceptable as shown below:

The red mark clearly says that this value is not satisfactory it's almost (25%) lost where as the acceptable value is (1%).The reason for this high rate packet loss as explained above busty web traffic occupies all the queue size so queue is full and router start tail dropping and voice packets are dropped as they are without any priority. The maximum delay observed for this stream is (4319ms) i.e. (4.319sec) which is a wild figure as its acceptable value is (150ms one way).

The graph shown below is kept after seconds behind the first captures by calling from head office to the branch office with the same setup, the jitter here first hit the peak value of (278) but it gradually decreases and at the end of the stream it obtain an acceptable value. The reason for this normalization of jitter value (although not acceptable) is that as soon as some busty web traffic went off from the queues of routers as previously it occupied all the queue size, voice traffic got some of there flows in and jitter value declined and an average value of jitter is almost like (75-80ms) that is also an unacceptable peak as in the case of above graph. But the packet loss has improved from the (25%) to (13.4%).

The maximum delay (max delta) observed during this call is about (4.319 sec) which is also an unacceptable rate. The delay can be calculated from the following formula as explained in RFC (3550). If (Si) is the RTP timestamp from packet (i), and (Ri) is the time of arrival in RTP timestamp units for packet (i), then for two packets (i) and (j), (D)is expressed as:

D(i,j) = (Rj - Ri) - (Sj - Si) = (Rj - Sj) - (Ri - Si)

4.1.2 Test Topology-2 (Complex Network)

This test topology is same as the network diagram for FR shown in the Figure-A.In order to examine the different behaviors of voice characteristics like jitter, delay and packet loss different test benches were designed to analyze the results critically. The major variable of this test bench are:

4.1.2.1 Test Bench-1

This test bench comprises of following criteria:

The results captured at branch office wireshark PC with the above mentioned test bench criteria. The jitter graph again shows unacceptable results as it hits the peak value of (50-51ms).As web traffic across the network occupy the queue size remarkably thus no enough space for the voice traffic. Although voice traffic got some streams in queue so therefore less packet loss is observed

as shown below it is (.05%) only, which is acceptable but at the same time max delta(delay) for this steam of traffic is about 672ms ,which indicates that although voice traffic got queued on router buffer but without any priority on the voice traffic it got delayed.

4.1.2.2 Test Bench-2

In test bench-1 wireshark PC is kept in the branch office between the CME and PC with CIPC so we can only get the traffic of the softphone across the network. As analog phone is connected to the CME and its traffic cannot be mirrored with this setup.So in order to monitor the traffic of both the phones a new setup is design as shown below:

Following graph highlight the jitter it strikes the peak value of (35ms) but an average jitter is something between (20 to 30ms) which is not bad with default fair queue.

But the packet loss value is much more then acceptable threshold it is (27%) means without any queuing policy on the router most of the packets are lost as queue become full with other traffic and may be some of them are lost in the network so therefore caller on the other end cannot understand the conversation as packet are lost.

The max delta (delay) value is (495ms) means there in a lot of delay during the conversation and listener on the other end get delayed voice packets. As the jitter value is not fluctuating much so there is a constant delay in the conversation.

4.1.2.3 Test Bench-3

This test bench is a special one as there are two >different Codecs on two different calls are implemented and results are also bit interesting. The calls are made before downloading the files from web server.

The jitter graph below shows acceptable value that is under (30ms),

The value of packet loss and max delta (42ms) is also acceptable for this stream as codec (G.711) take (64kbps) per call it occupies the bandwidth and therefore voice quality is better ,secondly G729 only take (8kbps) for its call so the total consumed bandwidth by these codec is (72 kbps).Since the calls are made before downloading starts so maximum bandwidth is captured by voice calls and there flows are in queue buffers before the web traffic flows thus getting there flows out 1^st as there is no priority applied on interface and by default fair queue operate its flow.

4.1.2.4 Test Bench-4(QoS Implementation)

As we have seen in all the above analysis graphs no QoS is implemented and we are getting remarkable values of jitter, packet loss and delay, now in this test bench CBWFQ is implemented:

The jitter graphs shown below fluctuate between (20-40ms) and if we calculate the average we will get (30ms) jitter,

4.1.2.5 Test Bench-5(QoS Implementation)

The results observed after the implementation of LLQ are very pleasant ,as the capture was taken at the branch office the jitter value is constant between (2.7ms) during the whole communication means it is getting the minimum delay in packets as only 24 ms one steam delay is observed which is much better then the acceptable value.

The reason for this minimum delay is strict priority for voice traffic which is (50%) of the bandwidth.

Tests for Wireless (QoS Implementation)

With CBWFQ

The above values of packet loss is (.32%) which is acceptable and max delay is (.087) sec which is also acceptable value, since voice is getting most of the bandwidth so therefore less packet loss is observed and as it got priority on other traffic based on the bandwidth assigned in CBWFQ .

The below graphs is delay graph observed at branch office it shows almost constant delay of (25ms) during the complete stream of voice.

LLQ Implementation:

The graphs below shows the jitter value when LLQ is implemented the jitter value is consistently fluctuating between (2 to 4ms),as priority bandwidth of (50) is assigned to the voice traffic so it is getting priority over all other traffic and therefore least jitter value is observed.

The below value of packet loss is (.04%) and max delay is (.066sec) both are acceptable.

CHAPTER NO.5

DISCUSSION

As the name of this section indicate clearly that this section of the report is dedicated to the discussion of the results obtained in the last results section.

Before going into the detailed discussion of test topologies and test bench design some important things need to be clearly addressed i.e. if we breakup the above section into two network parts then it will be easy to understand the test results detail debate.

• Simple Network

• Complex Network

  • Frame Relay

  • Wireless

5.1 Discussion of Simple Network:

Here simple network setup only include the test related to simple frame relay network with (one frame relay switch) connected to one CME on each side with one phone each. The results of jitter, packet loss and delay are wild for this setup. If we roughly draw a graph of the peak and lower values of jitter result obtained in the first result of simple network we will come up with something like showed in following graphs:

This graph shows clearly that jitter hits almost (340) as its lower value is 10ms during that voice stream. The reason for this peak value of jitter is heavy burst of web traffic flowing across the network before making the call. So when the call is made the voice traffic didn't find the space in the queue buffer and most of the packets are lost (tail drop) and secondly there is no congestion mechanism(queuing) on the router as by default (fair queue)is on the interfaces of router. Fair queue didn't differentiate or guarantees the bandwidth and delay thus web traffic flows occupy the full queue buffer size resulting packet loss and as the flows of voice traffic got some space in buffer behind the web traffic resulting maximum delay.

The second result is almost similar to the first one but this is much interesting as it was taken nearly after the first result. In this result although jitter and packet loss values are not acceptable but peak value of jitter is decreasing gradually as voice flows got some spaces in the queue buffer previously occupied by the busty web traffic since voice is getting through so tail drop also decreases at the queue buffers resulting low value of packet loss.

5.2 Discussion of Complex Network:

As explained above it is divided into two parts one for FR and other for wireless. So first talk about the frame relay. Since the network design is a bit changed by putting an extra switch between the two routers namely (Enterprise Edges and the CMEs) in head office and in the branch office. This change will now capture the traffic flow from the analog phones on both ends.

First test bench shows the jitter peak value of (50-51ms), reason for this value is web traffic. As delay is also very high in this test since web traffic across the network occupy the queue buffer remarkably thus leaving less queue buffer space for the voice traffic and voice traffic has to wait till the queue finishes sending the web flows from its interface then only giving chance to voice traffic to go out of interface and fair queue also didn't provide any guarantee for delay sensitive traffic resulting delay and packet loss of voice streams.

In second test bench G711 codec is used which take (64kbps) bandwidth for one call, thus giving good voice quality as compared to G729 which took (8kbps). But in this test packet loss value is much more (27%) then acceptable threshold means without any queuing policy i.e. congestion management we cannot guarantee the voice traffic and it is much needed. Most of the packets are lost as queue become full with other traffic and sometimes packets are lost in the network so therefore caller on the other end cannot understand the conversation as packets are lost.

Third test bench is very interesting; it is specially designed and tested to analyze the different factors of traffic. As said interesting its results are also very appealing The value of packet loss and delay (42ms) is also acceptable for this stream as codec (G.711) take (64kbps) per call it occupies the bandwidth and therefore voice quality is better ,secondly G729 only take (8kbps) for its call so the total consumed bandwidth by these codecs is (72 kbps).Since the calls are made before downloading starts so maximum bandwidth have been captured by voice calls and there flows are in queue buffers thus getting there flows out 1^st as there is no priority applied on interface and by default fair queue operate.

The following trend is observed after getting the average values of Jitter, packet loss and Delay for the three queuing techniques when test is done on the frame relay complex design .The three scattered lines shows there trends where fair queue is more scattered and give a straight line ,then CBWFQ gives a bit flatter trend but LLQ give good trend result almost flat trend. So it is obvious that from the series of results compiled LLQ is giving acceptable results for voice.

The following trend is observed after getting the average values of Jitter, packet loss and Delay in the wireless design for CBWFQ and LLQ.This shows that LLQ value better then the CBWFQ as it shows flatter trend then the CBWFQ.

The following trend shows the CBWFQ trend of Frame relay and wireless. It clearly shows that the results of (CBWFQ for wireless is much better then that of the frame relay based on the jitter, packetloss and delay.

The following trend shows the comparison of LLQ for both frame relay and wireless it shows that LLQ for the frame relay is better then that of the wireless.

CHAPTER NO.6

CONCLUSION

Did we think a decade ago that we can put voice transmission in packets and squeeze it to 8kb and still understand at receiving end? It just looks like a dream, an incredible dream which entirely revolutionize the telecommunication industry. It's not only that but the voice conduction traveled with the data traffic and rearranged by the vigorous devices at the recipient's side.

Enterprise are now designing there VoIP infrastructures along with the legacy networks because they are cost effective ,it increases revenues and profits with new applications, such as video calling, unified messaging, and web-enabled multimedia call centers. Giving more features voice mail, caller ID, call conferencing, call forwarding and unlimited long distance and they can be integrated with PC software>.

Problems with VoIP are its quality and reliability. During the implementation of this project these things are kept into consideration and quality of service is design and implemented for WAN setups. As Quality of Service (QoS) in VoIP is usually called energy of the voice show. The emphasis is given on Congestion Mechanism and (CBWFQ and LLQ) are implemented that offers prioritization over normal data traffic during the time of congestion and this has been observed. Here I would like to mention some limitations of the project, time constraint is the biggest limitation for me as I have to research and implement a few things secondly I love to do this project on Cisco Call Manager but it is not available in LAB.

Further Work:

Although I have achieved the entire objectives but still there are few things I didn't implement due to time limitation. So I would like to recommend them for the future enhancement.

• Test should be done on Cisco Call Manager.

• Header compression techniques should be implemented in QoS.

• Sophisticated Cisco IP phones should be used for test.

• A comparison and analyzation of VoIP should be done on some new technologiesCable, MPLS VPN and Metro Ethernet>.

REFERENCES

Ranjbar, A. (2008) CCNP ONT official exam certification guide Pearson Education ISBN: 8131714063

Tim, S(2006) End-to-End QoS Network Design Cisco Press ISBN: 1-58705-176-1

Wendell, Michael J. & Cavanaugh(2005) IP Telephony Self-Study:Cisco QOS Exam Certification Guide Second Edition Cisco Press ISBN: 1-58720-124-0

Kevin Wallace(2005)Voice over IP First-Step Cisco PressISBN: 1-58720-156-9

Kevin Wallace(2005Authorized Self-Study Guide Cisco Voice over IP (CVoice) Cisco Press ISBN: 1-58705-262-8

Sandy, G.. (2007) Cisco IP Telephony QoS Design Guide ISBN: 1-35875-365-2

Scott, E. & Roth,H(2008) CCNP ONT Portable Command Guide 1-58720-185-2

M,Matthias.(May,2008) Telephony Worldwide Market Share and Forecasts[Online] Available at: http://www.infonetics.com/pr/2008/ms08.pbx.1q08.nr.asp

IPv6 Foundation RFC 1889 http://www.ietf.org/rfc/rfc2460.txt?number=1889

IPv4 Foundation RFC 791 [online] Available at: http://www.ietf.org/rfc/rfc0791.txt?number=791

IPv6 Foundation RFC 2460 [online] available at: http://www.ietf.org/rfc/rfc2460.txt?number=2460

SIP Signaling RFC 3261[online] Available at> http://www.ietf.org/rfc/rfc3261.txt?number=3261

RTP Media RFC 3550[online] Available at> http://www.ietf.org/rfc/rfc3550.txt?number=3550

Hersent,O & Jean-Pierre Petit, IP Telephony, Packet-based multimedia communications systems, 2000

Mark A. Miller, P.E., Voice Over IP, Strategic for the Converged Network, 2000.

Forester Research, North American And European SMB VoIP Technologies Adoption' online /results.jsp?N=0&Ntk=MainSearch&Ntx=mode+MatchAllPartial&s=1&Ntt=voip

IT Manager's Journal, http://eyeonit.itmanagersjournal.com/eyeonit/

www.frforum.com

www.cisco.com

Sign up and be the first to receive our latest offers: