Integration of WLANS, PAN, LAN and GSM in Hmanets
Disclaimer: This dissertation has been submitted by a student. This is not an example of the work written by our professional dissertation writers. You can view samples of our professional work here.
Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.
INTERWORKING ISSUES IN INTEGRATION OF WLANS, PAN, LAN AND GSM IN HMANETS
KEY TO SYMBOLS OR ABBREVIATIONS
AP Access Points
AMASS Architecture for Mobile Ad-hoc Systems and Services
AODV Ad Hoc on Demand Distance Vector Routing
BS Base Station
BNEP Bluetooth Network Encapsulation Protocol
CDMA Code-Division Multiple Access
CGSR Cluster-head Gateway Switch Routing
CSMA/CA Carrier Sense Multiple Access with Collision Avoidance
CTS Clear to Send
DBTMA Dual Tone Multiple Access
DSDV Destination Sequenced Distance Vector Routing
DSR Dynamic Source Routing
GEO-TORA Geographical Temporally Ordered Routing Algorithm
GPRS General Packet Radio Service
GPS Global Positioning System
GRDL Grid Resource Description Language
GSM Global System for Mobile Communication
HF High Frequency
HMANET Heterogeneous Mobile Ad Hoc Network
HOLSR Hierarchical Optimized Link State Routing
IP Internet Protocol
LBR Location Based Routing
LLC Logical Link Control
MAC Medium Access Control
MACA Multi Hop Collision Avoidance
MACAW Medium Access Protocol for Wireless LAN
MAN Metropolitan Area Network
MANET Mobile Ad Hoc Network
MPR Multipoint Relays
NAT Network Address Translation
NFS Network File System
OLSR Optimized Link State Routing
OSI Open Systems Interconnection
PDA Personal Digital Assistant
QoS Quality of Service
RREP Route Reply
RREQ Route Request
RERR Route Error
SCTP Stream Control Transmission Protocol
SDR Software-Defined Radio
TBRPF Topology Broadcast Based on Reverse Path Forwarding
TC Topology Control
TCP Transmission Control Protocol
TDMA Time Division Multiple Access
TORA Temporally Ordered Routing Algorithm
VHF Very High Frequency
WAN Wide Area Network
WLAN Wireless Local Area Networks
WPAN Wireless Personal Area Network
WSDL Web Services Description Language
WSN Wireless Sensor Network
ZRP Zone Routing Protocol
Recent developments in wireless communications have taken possible applications from simple voice services in early cellular networks to newer integrated data applications. IEEE 802.11 family i.e. Wireless Local Area Networks (WLANs) have become popular for allowing low cost data transmissions . The most common and approachable places, such as airports, hotels, shopping places, university campuses and homes have been provided with WLAN Access Points (AP) which provide hotspot connectivity . The future advances in modern radios like Software-Defined Radio (SDR) and cognitive radio technologies will surely facilitate the need of multi-mode, multi-interface and multi-band communication devices. This heterogeneous networking paradigm will certainly enable a user to enjoy better service quality, ease of use and mobility, while keeping in view the application needs and types of available access networks e.g. cellular network, WLAN, wireless personal area network (WPAN) etc.
1.1 Mobile Ad hoc Networks
The Mobile Ad hoc Network (MANET) is a network formed by mobile wireless hosts without (necessarily) using a pre-existing infrastructure and the routes between these hosts may potentially contain multiple communication hops . The autonomous nature of participating mobile nodes enables MANETs to have dynamic and frequently changing network topology. The nodes are self-organizing and behave as routers. The ease and speed of deployment and decreased dependence on infrastructure have made ad hoc networks popular within very short span of time. MANET variations include Personal area networking (e.g. cell phone, laptop, ear phone), Military environments (e.g. soldiers, tanks, aircraft), Civilian environments (e.g. cab network, meeting rooms, sports stadiums), and foremost Emergency operations (e.g. search-and-rescue, policing and fire fighting). MANET's rapid deployment, ease of use and subsequent properties make them a hot choice for many important applications.
1.1.1 Resource Sharing
One of the intended aspects of MANETs is that it will facilitate the sharing of resources. These include both technical and information resources. Technical resources like bandwidth, Quality of Service (QoS), computational power, storage capacity and information resources include any kind of data from databases. Resource sharing among mobile devices require the devices to agree on communication protocols without the existence of any dedicated servers.
1.1.2 Coordination System
Mechanisms that enable the sharing of resources between different mobile devices, i.e. different coordination system is necessary for sharing dissimilar resources. Examples of such mechanisms are Samba, Network File System (NFS) for sharing disk space and the distributed dot net client for sharing processor cycles.
1.1.3 Trust Establishment
Before nodes start sharing any resource, they demand a certain amount of trust between them or systems with which they share resources. The level of trust depends on the kind of information or resources that is to be shared. For instance, sharing processor cycles require less trust than the sharing of personal information. Similarly, sharing of profit-making or highly sensitive information can require another level of trust establishment. There are systems currently in operation that can provide a certain amount of trust like the public key infrastructure that makes use of certificates.
1.1.4 Node Discovery
Before any node starts communicating with other node, that node must be discovered. When a node enters the network, it has to be capable of communicating to the other nodes about its capabilities e.g. it is a Personal Digital Assistant (PDA) and it has a camera, Global Positioning System (GPS) capabilities and Global System for Mobile Communication (GSM) capabilities etc. When a node is detected, other users can send a query to the new device to find out what it has to offer. Commercial service providers can advertise the resources they have to offer through Internet Protocol (IP) multicasts. There is a myriad of standards that include resource description protocols like Standards Grid Resource Description Language (GRDL), the Web Services Description Language (WSDL) for telling all offer devices a way to describe and publish their specific resources and needs. There are also various different systems currently available that can gather these resource descriptions and structure them for other devices to use.
1.1.5 Resource description
For any device to be able to use any resource, a way to identify and describe the resource has to be agreed upon by all available devices. If, for instance, storage capacity is to be shared, it first has to be clear what the capacity of each device is, and, what the storage need is. Although there are techniques to describe certain resources but not one technique that is able to provide this service for all resources. The available techniques combined, however, cover most of what is needed.
1.2 MANET Classifications
Mobile Ad hoc Networks are usually categorized as Homogeneous MANETs and Heterogeneous MANETs.
1.2.1 Homogeneous MANETs
When MANETs operate in fully Symmetric Environment whereby all nodes posses identical capabilities in terms of battery, processing powers, responsibilities and hardware and software capabilities, thus having no diversity, the network is Homogenous MANET.
1.2.2 Heterogeneous MANETs
In certain environments, mobile nodes may have asymmetric capabilities in terms of transmission ranges, Medium Access Control (MAC), battery life, processing powers, speed of movement and software variations etc. Mobility rate may also differ in ad hoc networks due to varying traffic characteristics, transmission ranges, reliability requirements and communication needs. Similarly, addressing and traffic flows like host-based addressing, content-based addressing or capability-based addressing patterns may be defined in certain scenarios; for example, people sitting at an airport lounge, metro taxi cabs, sportsmen playing and military movements etc
Homogenous MANETs do not allow for the heterogeneity in the network, which is seriously required in many scenarios, for instance, in a military battlefield network, where soldiers usually carry light portable wireless devices and more powerful equipment like High Frequency (HF) or Very High Frequency (VHF) is installed on vehicles. So, heterogeneous mobile nodes may co-exist in a single ad hoc network making it a Heterogeneous MANET.
1.3 Criterion for Heterogeneous MANETs
The integration of different communication networks like cellular networks, WLANs, and MANETs is not straightforward due to various communication scenarios, different interface capabilities and dynamic mobility patterns of mobile nodes. This exhibits many possible application scenarios where devices may unexpectedly interact, create and receive random data streams (video and music etc), or request different network services. The drawback is that each network type typically uses its own protocol stack especially in the case of medium access. In fact, frequency allocation becomes more complicated since different wireless technologies like IEEE 802.11 a and IEEE 802.15.4 may possibly operate in the same frequency band, which makes coexistence mechanisms increasingly important. A heterogeneous MANET paradigm needs to be capable of providing subsequent characteristics.
The network should be capable of providing seamless end-to-end communication among mobile nodes i.e. the MANET user must not be informed about the route followed or network interfaces traversed by a communication session .
Integration among dissimilar communication networks must facilitate mobile nodes via some mobility management framework that can manage flow of information through different medium access techniques .
Most of the IP based networks consider each communication interface as an independent network device running under its own protocol stack . However, this mechanism makes it difficult to remember destinations by IP addresses. So, there must be some mechanism similar to domain name service to recognize mobile nodes with more logical and easy to remember names.
Various configuration options like network ID, willingness for cooperative communications, desired mobility level and intended services shall be provided to mobile users for their convenience .
1.3.5 End to End security
Integration between various networks and data transfer over multiple wireless hops can even expose data to malicious nodes. Security mechanisms must take care of end-to-end data security as well as route security .
1.3.6 Transmission Power and interference of Nodes
MANET routing protocols must take care of issues arising due to various communication ranges like communication gray zones  and issues arising because of various communication technologies like Bluetooth and WLAN working in same frequency band .
1.3.7 Utilization of Resources
In heterogeneous networking paradigm, there may arise situations where some or most of the mobile nodes are installed with different kind of resources. For example, there may be some nodes installed with location monitoring devices like GPS. Now it is the responsibility of routing protocol to benefit from such capabilities in order to facilitate location aided routing and similar services .
1.4 Problem Statement
Current research efforts in mobile ad hoc networks are mainly converging towards inclusion of dissimilar communication technologies like IEEE 802.11  and IEEE 802.15.4  to a single mobile ad hoc network. Integrations of different networks like Wide Area Networks WANs (1G, 2G, 2.5G, 3G) and Metropolitan Area Networks MANs (IEEE 802.16) wherein users can access the system through a fixed base station (BS) or AP connected to a wired infrastructure in single hop fashion are also extending towards multihop communication environment using the new and revolutionary paradigm of a mobile ad hoc networks (MANETs), in which nodes constituting MANET serve as routers.
Comprehensive research efforts have been done to address the issues related to infrastructure-less multihop communications among nodes installed with dissimilar communication capabilities [3, 11, 12, 13]. However, an investigation needs to be made in order to analyze and address the issues arising from such integrations. Such problems relate to both end user's convenience (For example, remembering each destination with its IP address is a cumbersome job specially when every destination may carry multiple IP addresses and any communication interface may optionally be connected or disconnected) as well as network's performance; for example, routing to the best possible interface when there are multiple interfaces installed at destination. Likewise, optimized neighborhood sensing and position based routing can help to improve heterogeneous ad hoc network's performance and scalability.
The objective of this thesis is to study the integration of different technologies like WPAN, WLAN and GSM having different capabilities and protocol stacks to mobile ad hoc networks. Performance improvement issues relating to network configuration, human understandable naming mechanism and sophisticated location aided routing mechanisms will also be discussed and evaluated on an actual ad hoc network testbed.
1.6 Thesis Organization
Chapter 2 describes the different design and technological challenges arising from integration of multiple communication interfaces. Chapter 3 includes an overview of famous heterogeneous routing protocols architectures, interworking issues encountered, the limitation and solution suggested. Chapter 4 specifically discusses the adopted solution. Chapter 5 presents the details about solution implementation, protocol evaluation testbed, proposed test cases for evaluation of the proposed mechanisms and results obtained, whereas; chapter 6 concludes the research work.
2 LITERATURE REVIEW
The literature available on heterogeneous MANETs has suggested different combination of access technologies but no comprehensive solution comprising of maximum access technologies has been suggested yet. Some of the suggested techniques will be discussed in succeeding paragraphs.
2.2 Service Architecture for Heterogeneous IP Networks
It was presented by Joe C. Chan and Doan B. . This proposal is presented to resolve two main issues i.e. universal connectivity and MANET location management in heterogeneous networks. The new architecture suggested for Mobile Ad-hoc Systems and Services (AMASS) introduces a new abstraction layer called Mobile P2P overlay in order to cater for the problems such as transparency, dynamic routing, unique addressing, association, and application independence. Mobile users can associate local resources from neighboring devices, build wireless on-demand systems which is independent of location, hardware devices, networking technology and infrastructure availability. Five key design considerations considered were Mobile Peer-to-Peer Overlay, Internet Interworking, Intelligent Overlay Routing, Infrastructure-free Positioning and Application Layer Mobility. Three enhanced mobility models offered in this approach are Personal Mobility (using different IP devices while keeping the same address), Session Mobility (keeping the same session while changing IP devices) and Service Mobility (keeping personal services while moving between networks).
The architecture is built on a peer-to-peer communication model to integrate MANETs seamlessly into heterogeneous IP networks. Mobile Peer-to-Peer System(P2P) is a distributed Middleware addresses the demand of direct communication needs by creating spontaneous community. Whenever the Mobile P2P system has global connectivity, it works with its peer system and other applications systems by generic P2P signaling. It consists of Ad-hoc Network layer and Mobile P2P Overlay. The former layer includes wireless hardware and MANET routing software offering homogeneous connectivity among nodes with same wireless interfaces. These nodes act as a router forwarding traffic toward its destination. The later layer includes the following core services: (i) Membership Services offers single sign-on, naming, profile and identity features; (ii) Discovery Services for peer/resource discovery and caching; (iii) Communication Services for Internet interworking, intelligent routing, session control, presence and service delivery; (iv) Location Services for infrastructure-free positioning, and user mobility management functions; (v) Adaptation Services for application and network services adaptation.
Members of the Mobile P2P system should first sign-in a “common group” with their exclusive name and password. Some stationary nodes may also join to offer its resources such as Internet connection, printer, video conferencing. Whenever these client devices are within range of each other, they would work together as a team leading to a wireless adhoc service community where local resources could be shared by individual at its will. These members will then be available by intimating their capabilities and location information to the central location server. Information regarding physical location is also essential to offer spatial locality relationships and enable mobile content customization.
The results which were achieved through this process can be summarized as first, it maximizes the synergies of MANETs and P2P for building wireless on-demand systems and services. MANETs provide dynamic physical connectivity while P2P offers dynamic associations of entities (users, devices, and services) for direct resources sharing. Second, its Mobile P2P overlay unites mobility, user-centric connectivity, and services for universal communications. This allow dynamic service adaptations pertinent to user location, application requirements, and network environments. Third, it presents a flexible network structure stimulating fixed and wireless networks convergence. The result is an “Integrated Mobile Internet” which makes our future environment lot better.
2.3 Transparent Heterogeneous Mobile Ad hoc Networks
The idea was suggested by Patrick Stuedi and Gustavo Alonso. The paper discussed that performance issues in a personal area network (PAN) or wireless sensor network (WSN) may have less priority than an office network. In contrast, battery life and low cost is vital to PANs and WSN while most probably it is not an issue in an office network. Consider a scenario where in a certain university campus the students carry variety of personal devices like mobile phones, PDA or laptops equipped with different communication technologies tailored to their capabilities. The mobile phones or PDA will be using Bluetooth whereas laptops have 802.11 as well as a Bluetooth interface built in technology. Ubiquitously combining all these devices into one mobile ad hoc network could invite new applications and services such as location based services or VoIP. So there may be an occasion where a personal device of one particular PAN might communicate with a personal device of another PAN in a multi-hop fashion with the underlying MAC scheme changing per hop.
In this scenario two issues needs to be solved i.e. broadcast emulation and handover. Broadcast emulation is not directly supported in Bluetooth (nor on nodes comprising both Bluetooth and 802.11). Handover is an issue because, in the case of heterogeneous MANETs, a handover might include a change in how the medium is accessed. A handover can be caused by node mobility, a change in user preferences (where due to energy constraints the user chooses to use Bluetooth instead of 802.11), or performance reasons. 
Any device or node supporting multi interface though having different protocol stack will be specific to the interface at lower level. This characteristic will deteriorate the ability of a device to switch from one network to the other. The objective of such network is to provide an end-to-end communication abstraction that hides heterogeneity. The different possible design differs from each other with regard to application transparency, performance and mobility. There is another issue of handover which includes route changes as well as MAC switching. In principle, there are three possible scenarios
2.3.1 Horizontal Handover
The horizontal handover between the participating nodes take place when the route changes and underlying MAC technology remains the same.
2.3.2 Vertical Handover
The route does not change but the given neighbor is now reached through a new physical interface.
2.3.3 Diagonal Handover
The diagonal handover takes when the MAC technology and route between the participating nodes change simultaneously.
To address all these issues an IP based heterogeneous mobile ad hoc test bed using Bluetooth and IEEE 802.11 that implements a virtual interface approach as the end-to-end abstraction is presented.
2.4 Stream Control Transmission Protocol
Another approach presented by R. Stewart, Q. Xie, and K. Morneault is  Stream Control Transmission Protocol, a transport protocol defined by the IETF providing similar services to TCP. It ensures reliable, in-sequence delivery of messages. While TCP is byte-oriented, SCTP deals with framed messages. A major involvement of SCTP is its multi-homing support. One (or both) endpoints of a connection can consist of more than one IP addresses, enabling transparent fail-over between hosts or network cards.
Each interface could be separately cond and maintained (AODV-UU  e.g., supports multiple interfaces). This solution seems to be quite valuable in terms of performance since SCTP optimizes the transmission over multiple links. In fact, if one particular node can be reached through several interfaces, SCTP switches transmission from one interface to another after a predefined number of missing acknowledgements. Unfortunately, the solution lacks transparency. Applications running traditional unix sockets would have to be changed to use SCTP sockets instead. Another problem arises with the connection oriented nature of Bluetooth. In Bluetooth, interfaces appear and disappear dynamically depending on whether the connection to the specific node is currently up or down. Therefore, this is something that both the ad hoc routing protocol as well as SCTP would have to cope with.
2.5 Global connectivity for IPv6 Mobile Ad Hoc Networks
R. Wakikawa, J. T. Malinen, C. E. Perkins, A. Nilsson, and A. J. Tuominen, in 2003 through IETF Internet Draft, 2003 presented "Global connectivity for IPv6 Mobile Ad Hoc Networks," suggested one of the solutions for connecting heterogeneous MANETs. Before this work, the issue was solved by the traditional Internet model. But by adopting the approach presented by them the non structured MANETs were made to operate in structured environment, and inevitably limit the extent of flexibility and freedom that an evolving Mobile Internet can offer. Current mobile positioning and network mobility solutions are mainly infrastructure-driven which is contradictory to infrastructure-less MANETs. Without a flexible and user-centric network structure, existing solutions are generally insufficient to handle the dynamic and on-demand requirements of MANETs.
Heterogeneous MANET service architectures and routing protocols have been talked about and it is established that lots of enhancements need to be introduced to the heterogeneous ad hoc networks. First of all, different issues like IP addresses to hostname mapping and seamless communication need to be addressed. Secondly, ad hoc networks must seamlessly utilize all underlying interfaces. Finally, research efforts need to converge towards real-world network deployment as very few MANET service architectures have been evaluated on actual network testbeds.
3 INTEGRATION CHALLENGES
The invention of mobile devices like laptops, personal digital assistant (PDA), smart phones and other handheld gadgets having dissimilar communication interfaces smooth the progress of data transmissions without any predetermined infrastructure and centralized administration. . Such data transmissions can only be made on top of infrastructure-less networks composed of fully autonomous mobile nodes. But, these infrastructure-less networks do possess many complexities like dynamic and ever changing topology, heterogeneity in nodes, energy constraints, bandwidth constraints, limited security and scalability. However, there user-friendliness and rapid deployment make them an imperative part of 4th Generation (4G) architecture allowing the mobile users to communicate anytime, anywhere and with the help of any device.
3.2 Technological Challenges
The specialized nature of MANETs enforces many challenges for protocol design by incorporating changes in all layers of protocol stack . For instance, all the changes in link characteristics must be dealt with physical layer. MAC Layer should ensure fair channel access and avoidance of packet collision. Calculation of best possible routes among mobile autonomous nodes must be done by the network layer. Transport layer must modify its flow control mechanism to tolerate Packet loss and transmission delays arising because of wireless channel. The continuous making or breaking of connection due to nodes mobility be handled by application layer. These issues at each layer need to be handled effectively in order to smooth the transition from traditional network to advanced MANETs.
3.2.1 The Physical Layer
In heterogeneous MANETs there can be a node which may be able to access multiple networks simultaneously. If a node on one hand, is connected to a cellular network, and on the other hand it exists within the coverage area of an 802.11b AP, the network or the node should be able to switch between them. Moreover, in heterogeneous environment, different wireless technologies may operate in the same frequency band and it is significant that they coexist without degrading each other's performance. Therefore, techniques to reduce interference between nodes are important. For example, a node communicating with other nodes via multihop path may have lesser interference than a node communicating directly with AP. This is due to the attachment of increased number of nodes to the AP. Another way of reducing interference is Power control techniques applied in code-division multiple access (CDMA)-based cellular networks and MANET .
The research issues range from designing considerations to power control techniques include efficient design of nodes that can efficiently switch between different technologies and ensure higher data rates, development of Interference attenuation techniques between various wireless access technologies, modulation techniques and coding schemes that improve the performance of a given technology and frequency planning schemes for increasing the utilization of frequency spectrum
3.2.2 Link Layer
The data link layer can be divided into Logical Link Control (LLC) and Medium Access Control (MAC) layers. When a node needs to communicate to another node having cellular interface, it uses a centralized MAC access like Time Division Multiple Access (TDMA) or CDMA with a data rate upto 2.4 Mb/s. On the other hand, when a node communicates in 802.11 environments, it uses distributed random access scheme like Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) upto a data rate of 11 Mb/s. So, this difference of data rate is going to be one of the interworking considerations.
Due to dissimilar access technologies at intermediate hops, the performance of ad hoc networks deteriorates. The problems such as hidden and exposed terminals also limit the capacity of MANETs. The need of evolving mechanisms such as power control and power aware MAC protocols is mandatory to improve the performance of ad hoc networks. In a heterogeneous network, the cross-layer design may play a significant role in providing useful information to upper layers.
Another important issue to be considered at the link/MAC level is Security. Although, end-to-end security is the responsibility in the application layer, some wireless access technologies provide a certain level of security at the lower layers. Although the link and MAC layers in a multi-interface node can operate autonomously, but their operations have to be optimized to provide definite service to the upper layers. Some of the open issues include design of efficient link and MAC layer protocols to support QoS in Heterogonous MANET, channel administration schemes that consider different categories of traffic, and allow call blocking and handoff failure probabilities and security at Link / MAC layer.
3.2.3 Network Layer
The network layer needs to integrate all underlying communication interfaces; therefore, it is the most challenging task. The presence of nodes with multiple communication interfaces allow to have different physical and MAC layer technologies which need to be taken into account while dealing with an integrated routing process. But, the problem of MANETs such as frequent route changes due to mobility, higher communication overhead to learn and uphold valid routes, higher end-to-end delay and limited end-to-end capacity due to problems at the lower layers are main contributing factors in designing of routing process .
In order to reduce network control traffic, improve throughput and increase the range, the idea of integrating MANETs with infrastructure networks is evolved. Hence, mechanism to find gateways and correctly con IP addresses is required by such nodes in a MANET. The network layer has to find the best route between any source and destination pair. To define the best route, including number of hops, delay, throughput, signal strength, and so on several metrics can be used. Moreover, the network layer has to handle horizontal handoffs between the same technology and vertical handoffs between different technologies in a seamless manner. Several routing protocols have been presented for heterogeneous MANETs but the design of integrated and intelligent routing protocols is largely open for research with issues like development of routing capability in a heterogeneous environment that supports all communication possibilities between nodes forming MANET, scalability in multihop routing without significantly escalating the overhead and study of the impact of additional routing constraints (like co-channel interference, load balance, bandwidth), and requirements (services, speed, packet delay) needed by nodes and networks.
3.2.4 Transport Layer
In connection oriented transport session, as in case of Transmission Control Protocol (TCP), packet loss is assumed to occur due to congestion in the network. This assumptions leads to the performance degradation of TCP and factors such as channel errors, jitter and handoffs are overlooked. Moreover, in heterogeneous environments, the transport protocol has to handle the high delays involved in vertical hands off (while switching from one interface to another), server migration, and bandwidth aggregation .
Sometimes, a node changes its IP address when it needs to connect to another access network which may result in cancellation of ongoing communication sessions. Thus, transmission control mechanisms need to adoptively maintain the communication sessions previously held by the mobile node. One of the solutions for maintaining previous TCP sessions is the use of Mobile IP. But, it introduces relatively high delay so, newer and more optimized mechanisms are necessary. Similarly, due to presence of firewalls, server relocation support may be required when the node is not able to contact the original application server using the new access network address. Finally, the overlapping of coverage area between different access technologies can be exploited to improve the connection's bandwidth. However, the node must be able to judge the trade-off between achieved throughput, power consumption and cost before using multiple interfaces.
To enhance TCP performance in wireless networks, there has to be a mechanism for information sharing to tell the sender about the causes of the errors at the wireless links e.g. receiver centric approach . The suggested protocol keeps multiple states at the host according to the number of active interfaces. Then in a vertical handoff, the application can continue transmitting and receiving data in the old interface before the new connection is established. But, it should be noted that the advantages of a receiver-centric transport protocol are highlighted when the sender is in a fixed network. When both ends are connected to wireless access networks the errors can occur not only close to the receiver, but also near the last wireless link at the sender side. The main open problems at the transport level are to design a new transport protocol or adapt the existing protocols (mainly TCP) to take delays into account in vertical handoffs for end-to-end congestion control process, implementation of server migration technique without interrupting ongoing connections, and support bandwidth aggregation by taking advantage of the availability of multiple interfaces.
3.2.5 Application Layer
In a heterogeneous environment, all underlying complexity should be hidden and it should only allow access to the transport layer and network services as in the open systems interconnection (OSI) network model. The multiple access networks available in a specified location can also provide different types of application services to users. For example, WLAN APs placed at railway station can provide train information service. In this case, nodes installed with WLAN communication interfaces can discover and inform users about the availability of such service. In fixed networks, some particular nodes can be selected to store service availability information, while in MANETs decentralized service discovery schemes are required.
One of the problems is how to provide information about services available in the fixed network part participating in a MANET. Therefore, some kind of virtual service manager is needed that can filter relevant information. Due to multihop routing, the design of such schemes in the application layer becomes a critical issue to support collaboration in packet forwarding. Another fundamental problem is end-to-end security. In an adversarial environment, the heterogeneous network may suffer from various security threats that may degrade the efficiency of packet relaying, increase packet delivery latency, increase packet loss rate etc. Some of the open issues at the application layer include end-to-end security, service discovery mechanisms and billing / Charging mechanisms.
3.3 Design Issues
The classical wireless network has number of centralized entities like routers, base station, MSC etc. These entities coordinate the different functions of wireless nodes but MANETs lapse this infrastructure support. Since, ad hoc networks do not rely on any centralized entity; therefore, mobile nodes carry out these coordinations in a decentralized manner. Moreover, the connectivity between nodes can not be guaranteed due to wireless medium. For optimal MANET performance, the bandwidth utilizations must be minimized and transmission powers must be kept low to keep the battery available for longer time period. Keeping the transmission power minimum will lead to use of intermediate nodes for routing the information to the intended node. But at the same time the important feature of mobility in MANETs make the available routing information ever changing. This results in frequent exchange of control information between the nodes forming MANET. The frequent exchange will keep the wireless medium busy. In order to achieve high degree of reliability, much needed availability and most desired manageability the MANETs design / technological challenges are to be seen:
3.3.1 The Network Architecture
Network architecture is usually categorized as flat or hierarchal. Hierarchical network distributes the nodes in to different clusters depending upon technical or geographical requirements. Each cluster is headed by a single entry point, usually called cluster-head. Nodes in a particular cluster may be further divided into different hierarchies. On the other hand, flat architecture allows node to be at same hierarchal level i.e. zero tier hierarchical network is infact the flat architecture. Thus, the nodes residing in close proximities will establish communication easily whereas the routing problem becomes more aggravated for the nodes which are away from each other. Flat topology allows the ad hoc networks to locate multiple routes from source to destination. These routes usually introduce varying cumulative delays and data rates asking QoS based routing protocols. So, the time critical traffic adopts lesser delay route and rest of the traffic can follow normal route. On the other hand, routing in hierarchical networks is not optimal. There may be a possibility where two nodes may have less distance between them but due to involvement of cluster head nodes, the path followed is much longer than the actual distance. In addition, the use of one type of equipment in flat architecture offers no single point of failure as compared to cluster-heads in hierarchical network.
3.3.2 Routing Protocol
To ascertain the route from source to destination, it is necessary that at least the information regarding neighbor should be known. This requires frequent exchange of data between nodes, e.g. routing tables and route updates etc. When the size of network increases this volume of the information also increases to manifold. The problem gets further aggravated due to frequent mobility of MANET nodes, where due to increase in distance the link between the nodes may break much frequently. These frequent breakdowns increase the amount of topology control information resulting in higher amount of overhead. This exchange of topology control information between mobile nodes may even take a major portion of limited network capacity, further trimming down the proportion of actual traffic.
MANET protocols are categorized as either proactive or reactive . The proactive protocols are based on Link State protocols e.g. OLSR  where every node has complete details of the routes available from source to destination. This allows proactive protocol to immediately determine the route to any destination. However, the proactive topology broadcasts make these protocols a second choice for larger networks. On the other hand, the reactive protocols are based on flooding algorithms where route information is gathered by sending the request in a flooding manner to the nodes on demand only e.g. AODV  and DSR . Since the reactive protocols locate routes in, on demand fashion; so, they are not suited for real time communication. In view of the fact that MANETs have the major feature of mobility; therefore, continuous information updates regarding movement of nodes are required. Thus, cost of the network is much more in proactive protocol due to frequent exchange of topology information between each node in MANETs. So, there is need for a protocol which initiates the route request on-demand as well as has some information regarding routes to certain nodes. 2.1 depicts the routes available to node B. The results by using proactive, reactive, hybrid and location based routing protocols are summarized in Table 2.1.
As per Zone Routing Protocol (ZRP) , each node defines its area around is called node's zone. When the route is to be determined to the destination node not present in the node's zone, a request is generated to designated node situated at the border of its zone. This exchange of information between selected nodes results into less traffic as compared to real proactive protocol. Moreover the process of searching of nodes is due to presence of zone is much faster as compared to reactive protocol.
Another approach is Location Based Routing (LBR) in which the packets are forwarded to destination node's direction. But, performance of all of these protocols depends upon network conditions such as traffic load, number of nodes, mobility pattern and distance between nodes.
3.3.3 Medium Access Control
Since, MANET nodes operate in same channel; so, efficient channel utilization is highly desirable for efficient MANET operating. The ALOHA and Slotted ALOHA are initial MAC protocols. But, both of them failed due to their less channel utilization which is 18% and 36% respectively. They remained applicable where long propagation delays are acceptable like satellite communication. This leads to Carrier Sense Multiple Access (CSMA) protocol which has low probability of collision. But, due to inherent characteristic of MANETs where nodes move frequently, the Hidden Terminal Problem  exists therefore chance of collision remains there despite of using CSMA. Furthermore, in CSMA protocol the sensing of medium to be free is judged from transmitter but in case of collision at receiver forces the nodes to defer transmission and this problem is called as Exposed Terminal Problem. These two problems have made this scheme not valid for MANETs. 2.2 presents hidden terminal problem and exposed terminal problem.
The advanced protocols like Multi Hop Collision Avoidance (MACA)  and the Medium Access Protocol for Wireless LAN (MACAW)  use RTS / CTS (Request to Send / Clear to Send) mechanism. These RTS / CTS Dialog reserve the channel to avoid collision. The IEEE 802.11 wireless standard present the idea of Carrier Sense Multiple Access with Collision Avoidance (CSMA / CA) which solves the problems of initial protocols by sending positive acknowledgments to the source node and allows retransmission of unacknowledged packets.
But, all of these schemes do not address amicably the connectivity issue between source and destination nodes in large MANETs. In a highly mobile network if interfering nodes do not hear RTS / CTS dialog then large network capacity is wasted due to collision of packets. And if interfering nodes hear RTS / CTS dialog then the source or destination nodes defer their transmission and waste the capacity of MANETs as well.
Therefore the requirement is there to have protocol which not only provides the status of destination node but give the status of channel. This is more important where nodes in MANET, getting close to any node due to its mobility, do not interfere with already going on communication between the nodes. The Dual Tone Multiple Access (DBTMA)  suggests somewhat solution in which single common channel splits into Data channel and Control Channel. Moreover, two busy tones are assigned to control channel which is receive busy tone and transmit busy tone. Thus during the period when any of the tone (receive or transmit) is maintained between the source and destination nodes, the new incoming node refrain itself from getting the media access.
3.3.4 Service and Route Discovery
Since nodes keep on joining and disjoining the MANETs therefore the capabilities and services available with incoming nodes may not be known to the already existing nodes in MANETs. There can be two types of service offering mechanism. One is directory_less service and other is directory based service. The former is the mechanism in which some nodes request services from other nodes when needed or may be called in a reactive fashion. The other nodes announce their services in proactive fashion. On the other hand in the directory based services, nodes register the services they have with directory agents. These agents keep updated information as nodes join and disjoin MANET at their will. Due to this vary nature of MANETs current discovery services like UPnP could not do well in MANET environment.
But, still we require one comprehensive protocol for this type of services which caters for all requirements of MANETs like location of nodes, node profile, early notice for node movement etc. Moreover, when the environment becomes heterogeneous that is MANETs connected to fixed network or GSM then this protocol needs to understand the same type of services available with these type of network and evolve the mechanism to make these services available to MANET nodes.
MANET routing protocols run on top of network layer, so, for every communication interface, it is necessary to have an IP address attached to it. This becomes even more subtle when MANETs are attached with internet. However, this is not a convenient idea for an end user's prospect to locate some other mobile node with its IP address. Moreover, dynamic nature of ad hoc networks and autonomous nature of partaking nodes do not allow any user to have fixed IP address. Secondly, if a mobile node is installed with multiple communication interfaces, it needs to have multiple IP addresses which make it even more difficult to locate .
For a mobile node attached to fixed network, the problem of addressing is little cumbersome. Problem of addressing nodes joining and leaving the network can be dealt by using Mobile IP concept. Another solution could be having unique address with duplication address detection mechanism. Use of Network Address Translation (NAT) will facilitate the outgoing traffic from MANET to internet. But again the presence of multiple internet gateways create problem for NAT because in case of changing the gateway the IP address will change and session breaks.
Use of subnet in MANET so that all node in one particular network can have unique address may be one of the solutions. But this can only solve the problem of node being inside or outside the MANET. But there exists no solution for assigning unique prefixes to nodes to handle the different points of attachment to the internet. Therefore no common adopted solution exists currently. A need is felt to have IP address for routing only and having some other name configuration for addressing.
The presence of wireless medium in MANETs increases the security risks . In case of malicious node, it can come in between two nodes and listen to the information being shared. This is difficult to detect since attacker does not generate any traffic. Furthermore implementation of cryptographic mechanism is difficult in MANETs. Since, there is no centralized key distribution and certification authority.
But, there may be some nodes which become part of the network and do not cooperate in routing between different nodes. These types of nodes are called selfish nodes. Such nodes create much problem since MANETs rely on cooperative communications among mobile nodes. To cater such issues one of the solution could be to have billing system in vogue for MANETs where nodes are bound to cooperate. But this is subtle since in case of MANETs connected with internet. Then, only internet gateway will be billed making it difficult to realize which node to be billed or which node not to be. But there will be requirement of combination of both, i.e. trust establishment and billing mechanism .
The discussion in this chapter clearly suggests that efficient routing in an ad hoc network requires that the issues arising due to inclusion of multiple communication interfaces and dissimilar protocol stacks must be addressed at each layer of protocol stack. Such provisions and customizations can make ad hoc networks to perform well to support useful applications.
4 PROPOSED SOLUTION mechanisms
Integration of heterogeneous communication interfaces in a single ad hoc network poses many challenges on performance of MANETs and introduces a number of constraints and complexities to their design. This research addresses a number of issues like addressing, location based routing and optimized neighborhood sensing etc
Since, most of the existing MANET routing protocols are IP based routing protocols, so; every communication interface needs to have its own IP address . Thus, a mobile node installed with multiple interfaces must have multiple IP addresses attached to it. These addresses make it very difficult for MANET users to remember all IP addresses carried by a particular node. This becomes even difficult with the frequent inclusion or exclusion of interfaces. For instance, consider a heterogeneous ad hoc network as presented by 4.1. Now, node S wants to communicate to node D. Node D has two communication interfaces with IP addresses 192.168.1.3 and 192.168.1.4. Node S already has an active route to node D IP address 192.168.1.3. But, a route search to node D IP address 192.168.1.4 triggers a new route discovery, overburdening the network resources. Secondly, different routes to a particular node offer different service parameters and it's easier to find the best route to a destination by searching the destination itself rather than the IP address. Moreover, it's not convenient from an end user's perspective to remember all IP addresses for every intended destination especially when mobile nodes have dynamic address allocations.
Thus, it is intended to address every mobile node with its name rather than its IP address. For this purpose, AODV's Route Request (RREQ) and RREP messages are modified to to facilitate IP to name mappings. Destination IP field present in the RREQ message is modified to carry Destination Address as presented by 4.2.
(a) RREQ Structure
(b) Modified RREQ Structure
Similarly, Destination Name field is appended to AODV's RREP message. Since, RREP is always a unicast message, so it does not introduce any noticeable overhead to AODV routing protocol along with providing applicable convenience (IP address to name mapping). 4.3 presents the modifications introduces to the standard AODV RREP message.
(a) RREP Structure
(b) RREQ Structure
On receiving this RREP, the originator starts communicating with the destination using destination's IP address identified by the route discovery process. If a route break occurs, RERR for broken route is sent to originator. This RERR message carries AODV's default RERR structure. On receiving the RERR, originator consults the routing table to locate the destination name and initiates the route discovery process to the destination. 4.4 presents the flow of events for AODV's modified route discovery process.
4.3 Optimized Neighborhood Sensing
Mobility is one of the most desired attribute of mobile ad hoc networks. But, mobility directly affects the working of MANET routing protocols i.e. increase in mobility starts introducing performance penalties. At very high node mobility, routes set up by the reactive routing protocols may even become void even with-out use; thus, nullifying all efforts exerted to locate that path. Since, reactive protocols like AODV identify their live neighbors via regular beaconing (hello messages)  which usually requires higher response time e.g. it would require 6 seconds to identify a link break if a hello message is broadcast after two seconds and two missing hellos point to a link loss.
We tune AODV to adoptively settle on network's mobility level via frequent exchange to hello messages i.e. increase in participating node mobility with corresponding decrease in hello messages interval. This frequent exchange of hello messages will surely consume higher share of network traffic even increasing with the size of network. Moreover, partaking nodes are tuned to utilize hello messages in a more efficient manner, i.e. neighborhood sensing is performed only if a node is taking part in active communications or if that node is listed as a precursor to some other route.
4.4 Location Aided Routing
For a route discovery process, an AODV enabled mobile node broadcasts the RREQ to its neighbors in the transmission range. On receiving this RREQ, the neighbor nodes check if they are the intended destination. If so, a route reply is sent back and a route is established. Otherwise, the neighbor node acts as an intermediate node and rebroadcasts the route request to the nodes in their transmission range. This process continues until the destination node is found but this rebroadcast of control messages creates network congestion huge network congestions. This congestion and control traffic increases with the introduction of Heterogeneity in MANETs and internetworking of MANETs becomes more complex operation. If some or all of the participating nodes are installed with location estimation devices like GPS or location calculation algorithms , the routing protocol must be capable of using this location information in route discovery process. Such a modification will enable the route requests to flow only towards the intended destination using the location information available with each node. Such modifications can save network bandwidth, help to avoid congestion and enhance the network performance by minimizing the overhead in route request search and route establishment mechanisms.
4.4.1 Packet Modifications Made
Modifications are introduced to the AODV's route discovery mechanism. For this purpose, RREQ packet is modified to contain location information of the intended destination, distance value between source and destination and location information of originator to aid in RREQ forwarding. During route discovery process, this route request message is forwarded along with one extra field to contain the intended geocast region i.e. the area where intended receivers of the transmission are present.
4.4.2 Route Request Forwarding
On receiving the RREQ message, the neighbor nodes will extract the distance field form route the route request to compute the distance between its own location and destination location. This distance between intermediate node and destination node is compared with that of source and destination. The neighbor located closer to the destination will have somewhat small distance and can forward the RREQ towards the destination. On the other hand, if the distance of neighbor is relatively greater than the source indicates that this intermediate node is not situated closer to the destination so this node will discard the RREQ message. Thus, flooding of RREQ is curtailed. This helps in faster establishment of route with the destination since the intended direction of the destination is ascertained.
4.5 Interworking between MANET and GPRS
Currently GSM networks cover most of the geographical areas. Especially in urban environments, GSM networks cover almost every inch of land. It can be very useful to utilize GSM networks as an alternative when a particular destination can not be reached via MANET nodes. Moreover, this provision can be very useful to connect geographically displaced ad hoc networks. However, certain problems arise on interworking between MANET and General Packet Radio Service (GPRS) networks. First, GPRS networks are commercial networks offering volume based billing mechanisms. So, we cannot merely air hello messages over GPRS networks especially when it is not necessary because it will significantly increase the traffic volume over GPRS network. Secondly, message delivery through GPRS networks usually takes extended time (may be a few seconds) which usually result in route discovery time-outs at host or intermediate nodes. So, routes including GPRS hops are actually not considered by the route estimation module.
Thus, route discovery mechanisms and timeout configuration parameters are modified to accommodate routes comprising of GPRS hops. For this purpose, few alterations are added to default route mechanism. A route request is aired over GPRS interfaces only if it is desired by the user. Secondly, route timeout are extended to ten seconds for GPRS interfaces.
Heterogeneous ad hoc networks and issues arising from integration of dissimilar communication interfaces and capabilities are discussed and some improvements targeting user convenience and network performance are considered. First, an addressing mechanism is introduced to AODV routing protocol which adds to user convenience without introducing any routing overhead to AODV. Secondly, modifications are introduced to default AODV to benefit from the location estimation devices to lower the control traffic involved in route discovery and route maintenance mechanisms. Some configuration parameters and improvements are also proposed towards optimized neighborhood sensing and routing over GPRS interfaces.
5 RESULTS AND ANALYSIS
The proposed solution mechanisms address very important issues arising as a result of integrating multiple communication interfaces in a single mobile ad hoc network. These enhancements are incorporated to AODV's public implementation AODV-UU . The modified protocol is run and assessed on a real-world ad hoc network testbed composed of seven mobile nodes installed with multiple communication interfaces (Bluetooth, WLAN and Ethernet). 5.1 presents the testbed nodes, IP addresses and communication interfaces carried by each node. Moreover, every mobile node is assigned a name (A-G) to identify it in a more convenient way.
The hardware specifications of machines used for the testbed are presented in table 5.1. The evaluation testbed is composed of three laptop and four desktop machines running Fedora 7 (Linux Kernel 2.6.13) along with modified AODV. IPTables  are used to simulate mobility patterns i.e. to pretend as if a node has moved out of the coverage area. Wireshark is used for packet sniffing and performance monitoring.
Table 5.1 Testbed Hardware Specifications
Intel ® Pentium ® IV 3.0 GHz, 1 GB RAM
Running Fedora 7 (Linux Kernel 2.6.13)
D-Link ® DWL-G122
RTL8139 PCI Fast Ethernet NIC
Intel ® Pentium ® IV 2.4 GHz, 768 MB RAM
Running Fedora 7 (Linux Kernel 2.6.13)
D-Link ® DWL-G122
Billionton Class1 Bluetooth
Intel ® Pentium ® IV 2.4 GHz, 768 MB RAM
D-Link ® DWL-G122
RTL8139 PCI Fast Ethernet NIC
Intel ® Pentium ® IV 2.4 GHz, 768 MB RAM
D-Link ® DWL-G122
Intel ® Pentium ® IV 2.4 GHz, 768 MB RAM
D-Link ® DWL-G122
Billionton Class1 Bluetooth
Intel® Centrino Duo®2.4 GHz,512 MB RAM
RTL8139 PCI Fast Ethernet NIC Billionton Class1 Bluetooth
Intel ® Pentium ® M 1.6 GHz, 512 MB RAM
RTL8139 PCI Fast Ethernet NIC
Since, AODV is an IP based routing protocol running on top of IP layer. So, all communication interfaces submissive to IP V4 traffic can be taken in by AODV. This allows running AODV on Ethernet and WLAN interfaces. However, Bluetooth has a different protocol stack disagreeing to IP v4 traffic so AODV cannot be run on Bluetooth. This problem is resolved by running Bluetooth Network Encapsulation Protocol (BNEP)  on Bluetooth. BNEP hides the underlying master slave topology of Bluetooth by encapsulating the IP header with its own header; whereas, Bluetoth header is replaced by the IP header on receiver's side. This process emulates Bluetooth as Ethernet interface in the TCP/IP stack. BNEP protocol is presented in 5.2.
5.2 Test Scenarios
This section provides assessment scenarios and results obtained for the validation of proposed scheme. First, assessment scenarios for addressing mechanism are provided and then performance evaluations of location aided routing mechanisms are given.
For evaluation of proposed addressing mechanism, three test cases are defined as presented by 5.3, 5.4 and 5.5.
220.127.116.11 Test Case1
As a first test case, a route search for destination node F was made from source node G as depicted by -5.3. The route request was made using the destination name rather than its IP address. Proposed addressing mechanism successfully enabled AODV to locate a valid route (route 1) to the destination node F.
18.104.22.168 Test Case2
In the second test case, a route search was initiated from source node A to destination node E using the destination name rather than its IP address in the route request. Addressing enabled AODV was able to locate route 1 to destination the node E (IP address 192.168.3.5) which corresponds to the best route taken by AODV. Firewalling the NODE F with other nodes via IPTables resulted a route break to route-1. The suggested improvements to the AODV's route maintenance mechanism made possible for AODV to locate an alternate route (route 2) to destination node E which corresponds to a different IP address (192.168.2.5) at node E. 6.4 (a) and -6.4 (b) presents the route1 and route2 to destination node E.
22.214.171.124 Test Case3
In the third scenario, a route discovery was initiated from source node D to destination node F using destination's name, instead of its IP address. The proposed addressing mechanism enabled AODV to locate a valid route to destination IP address (192.168.1.6). Firewalling the source node D with its next hop B resulted in a route break to route 1. The proposed addressing mechanism enabled AODV to locate route 2 towards destination; although, the new route corresponds to a different IP address (192.168.3.6) at destination F. 5.5 (a) and 5.5 (b) presents the route-1 and route 2 taken by AODV's modified route discovery mechanisms.
Table 5.2 presents the results obtained in evaluation test cases. It is evident from obtained results that proposed addressing based route discovery mechanism performs equivalent to AODV's route discovery mechanism; whereas, Test-case2 route2 and Test-case3 route2 presents that proposed mechanism made possible for AODV's route maintenance algorithm to locate new route targeting different IP address at same destination which was not possible for conventional AODV.
Table 5.2: Test and Results
Proposed Naming Mechanism
Test Case 1
Test Case 2
Test Case 3
5.2.2 Location Aided Routing
For this test scenario, the evaluation testbed is established as presented by 5.6. The testbed is characterized by heterogeneous interfaces of different communication technologies (Wi-Fi, Bluetooth and Ethernet). In this experiment, we deliberately installed the interfaces in a way that switching became inevitable for the route discovery (RREQ) messages making it to traverses all different interfaces in order to reach the destination. 5.6 depicts the traversing of route request packets towards the destination. A ping request was sent from M2 node IP address 192.168.7.2 to LTZ node IP address 192.168.8.5. As M2 had only 1 communication interface (Bluetooth) so it was only able to reach M1 and M5 nodes in that plane. Now, on receiving the route request from M2, both M1 and M5 calculated the distances and because M5 was not in the forwarding zone so it discarded the RREQ. Whereas, M1 being inside the forwarding zone, forwarded the route discovery packets to its
Cite This Dissertation
To export a reference to this article please select a referencing stye below: