This research aims at improving the performance of resource reservation protocol in mobile networks. Today, more and more portable devices such as mobile phones and laptops can easily access the wireless network. Hence, more and more people choose the wireless devices due to the convenience they provide. When a large number of these devices are accessing the mobile networks, it could lead to service disruptions, packets loss, and accumulative delay. Therefore, the question of how to provide a stable quality of service has become an important research topic. Researchers focus on reservation approaches in order to solve the problems.
The resource reservation protocol (RSVP) (Braden, et al., 1997; Tschofenig, et al., 2006; Bingyi, and Yamin, 2004; Tseng, et al., 2005; Wroclawski, 1997; Zappala, et al., 1993; Ping, and Henning, 1999) was originally designed for fixed networks to reserve resources across the networks. It provides receiver-initiated setup for multicast and unicast data flow. It uses the PATH and RESV messages to setup the reservation between the sender and the receiver. In order to maintain the reservation state, the sender and the receiver need to periodically send the refresh Path and Resv messages through the networks. RSVP was also designed for mobile networks, today. The working process is similar to fixed networks. However, in mobile networks, when the mobile node moves from one location to another, the previous reservation path will be 'torn' down. The mobile node starts to reserve the new resources only after a new path is reestablished. As such, exceedingly large number of messages in RSVP will make the reservation very complicated and waste much resource. Hence, reservation protocols in mobile networks need to be efficient, adaptable, optimal, secure and simple.
There is various reservation protocols have been developed for mobile networks. The goal of these reservation protocols is to establish an available, stable and high quality of service in mobile networks. These protocols are categorized as Heavyweight Reservation Protocol and Lightweight Reservation Protocol. In the heavyweight reservation protocol, each router maintains a number of messages in order to store resource reservation and update the information when the network is changing. It consumes huge bandwidth. On the other hand, in the lightweight reservation protocol, it designs a small group of messages for reservation and updating. Lightweight reservation protocol incurs less overhead as compared to the heavyweight reservation protocol. The goal of these reservation protocols is to establish a stable and high quality of service in mobile networks.
In order to reduce unnecessary bandwidth consumption, the lightweight reservation protocol is used. Some example of the lightweight reservation protocol includes RSVP-TE (Resource Reservation Protocol -Traffic Engineering) (Awduche, et al., 2002; Rosen, 2000; Awduche, et al., 2001), Boomerang (Bergkvist, and Cselenyi, 1999), YESSIR (Yet another Sender Session Internet Reservations) Protocol (Awduche, Ping, Henning, 1999 and Agu, 1997); SMRP (Sender-initiated and Mobility-support Reservation Protocol) (Shangguan, et al., 2000), etc. These protocols are sender-initiated reservation protocols (sender is a key part), which consume less bandwidth and overhead when the mobile node needs to move.
Quality-of-Service (QoS) is a common problem in mobile networks. Such as the rapid growth of wireless network devices need to access wireless network easily and conveniently. Thus, whatever applications used need efficient services similar to that offered by the fixed network. Researchers have proposed using resource reservation in order to reduce the service delay-time and service disruption. Despite this, many other problems are encountered. These include: message overloading and congestion; patchy processing; wasted bandwidth; and degradation of the network performance. This is because wireless channels break down easily when the mobile node moves.
Reservation in mobile networks suffers from an inherent problem -- the difficulty in achieving a continuous path establishment. This problem occurs when the connection of the moving mobile node is disrupted intermittently. Mobile network is a multi-hop network wherein message is forwarded to the destination node through intermediate nodes. Frequent movement of a mobile node brings about much negative effects to the reservation establishment. It affects the overall performance of the network. Sender-initiated Mobility-support Reservation Protocol (SMRP) works better in mobile environment, and it is known as a lightweight resource reservation protocol. The sender plays a key role wherein it sends the path finding and path reservation request message to the receiver. The receiver responds by sending the final version of the reservation result to the sender. When the receiver receives the message, successfully, it sends an ECHO message to the sender. SMRP has a 'soft state' as well, but it is different from the 'soft state' of RSVP. In RSVP, the PATH message and RESV message need to be period refreshed in order to maintain the 'soft state'. Whereas in SMRP, when no data is being transmitted in the path, the sender only sends the reservation-RESV to refresh and maintain the reservation state. If data is being transmitted in the path, it does not send the reservation-RESV message. Hence, compared to the RSVP, the SMRP reduces the overhead messages by refreshing with the RESV message only, and the PATH message is not needed. Thus, the SMRP is preferred for implementation in mobile network, currently.
1.3 Statement of Problem
Sender-initiated Mobility-support Reservation Protocol (SMRP) is a lightweight reservation protocol, wherein the path-finding and path reservation are working together (Shangguan, et al 2000). In the mobile networks, when a mobile node changes location, the original reservation path is disrupted and service will be significantly degraded during the handover due to the lack of reserved resources in the new location.
Typically, handover procedure generates the route break and the packets will be discarded. The failed route produces much overhead due to lost packets being re-sent by the source node to reestablish a new path. Therefore, a lot of bandwidth is consumed, which results in higher overhead. If the mobile node moves drastically, the overhead increases rapidly. As a result, the quality of service would be significantly degraded.
In order to provide guaranteed service in mobile networks, the mobile node (MN) must re-establish a new reservation path from the new location to the source node as quickly as possible. Therefore, any delay caused by the handover procedure would be minimized. The COR (Crossover Router) scheme (ATM Forum 1997) was proposed as an extension to SMRP for mobile hosts, but it does not provide a smooth handover procedure for the reservation. It does not guarantee the quality of service when mobile node moves. Hence, it is very important to improve the overall network performance by designing a new network topology, which provides a smooth handover, and solve the unnecessary resending of data when the old reservation path fails.
Crossover router (COR) is a node where the old path and the new path meet. It includes the flow Session ID (SID), the Flow ID (FID), and the peer identification, which is changed if the route changes (Sanda and Fu, 2007). When there is a handover of MN, the guaranteed QoS will be affected. It is therefore, recommended that the resource reservation has to be made before the handover of MN, to guarantee continuous QoS. This is called Pre-Resouce Reservation. If a common path exists between a passive path and an active path, it is important to discover a crossover router where the paths meet. The crossover router cannot be applied for the pre-resource reservation because the FID and peer identification have not changed as the handover has not occurred yet (Sanda and Fu, 2007). Moreover, since the data traffic of the MN is transmitted on the common path, the reservation state on the common path should not be replaced with the new one.
The Crossover router discovery scheme (CRD) (Sen, et al., 2000) has been widely used because of the achievement of low handoff latency, and the signaling load making full use of the previous common path. Only the partial branches between a crossover router and the MN need to be re-established. However, when fast and frequent handoff occurs, the crossover router scheme may not work efficiently, and it is limited to the tree-based network topology.
ThePointer Forwarding (PF) scheme (Lee, Gwo-Chung et al 2001) introduces triangle routing or excessively long reservation path problem during several handoffs (Lee, Gwo-Chung et al 2001). The intermediate nodes in the unnecessarily long path may result in severe congestion as they require supporting more traffic than they can handle. It is also limited in the tree-based network topology, thus, when the tree topology is really high, there will be higher number of nodes. Therefore, handover occurs frequently. When many mobile nodes enter the network, which becomes 'exhausted' as there is too much resource due to the long-forwarding chain.
1.4 Statement of Objectives
This research is undertaken with the aim of improving the performance of the reservation protocol by using a hybrid scheme in mobile networks. The aim will be achieved by fulfilling the following objectives:
To study the Resource Reservation approaches in fixed and mobile networks, as well as the existing lightweight reservation approaches in mobile networks.
To model a hybrid scheme for the purpose of: reducing the reservation handover time delay, and packet drop probability; and increasing throughput.
To simulate the proposed hybrid model using NS2 to collect simulation data.
To evaluate the performance of the proposed hybrid model. This is done to compare the performance of the COR and COR-PF schemes, and to check whether the performance of the hybrid scheme is better than the COR scheme, according to our proposed performance metrics.
1.5 Proposed Solution
In this research, we address the problems by proposing the Crossover Router- Pointer Forwarding scheme, called the hybrid scheme, which combines the crossover router and the pointer forwarding scheme. We also propose a simulation study to evaluate the effect on performance of reservation handover delay, packet drop probability and throughput in mobile networks. Finally, we compare the performance of COR and hybrid scheme.
The Sender-initiated and Mobility-support Reservation Protocol (SMRP) was selected as a reference example of reservation protocol to evaluate the reservation handover delay, packet drop probability and throughput in the research. The SMRP protocol was chosen because of its simple feature, which is a lightweight resource reservation used in mobile networks. Compared to other reservation protocols, it has less processing overhead -- message overhead.
The proposed solution is implemented by using Network Simulator version 2 (NS2) (Chen, and Huang, 2000). It is an object-oriented, open source network simulation tool, which has many modules embedded in each node. NS2 is a free simulation tool, which provides many routing and transport protocols for fixed and mobile networks.
In this research, focus is on one problem -- testing the COR scheme and hybrid scheme with respect to handover delay, packet drop probability and throughput in mobile networks.
1.7 Thesis Organization
Figure 1-1 shows an overview of the various stages of the research. The dissertation has 7 chapters that are organized as follows:
Chapter 1 is an introduction to readers about background, motivations, objectives, scope. Background section presents the concept of resource reservation protocol and highlights the importance in the networks. Motivation section presents the performance degradation of the reservation protocol. Statement of Problem highlights the problem for our thesis. Scopes present the reason of choosing SMRP.
Chapter 2 presents the literature review. It discusses in details about the reservation in fixed network, example of Resource Reservation Protocol (RSVP). The following section presents the reservation protocols in mobile networks; example of RSVP, RSVP-RA, Mobile RSVP, and Hierarchical Mobile RSVP. In the last section, we describe the lightweight resource reservation approaches, example as RSVP-TE, Boomerang, and YESSIR (Yet another Sender Session Internet Reservations) Protocol and SMRP.
Chapter 3 presents the literature review of crossover router scheme and pointer forwarding scheme, it includes the definition, the functions of the schemes. Then a comparison has been made between these two schemes. In the end proposed solution of this research.
Chapter 4 models the solution which decreases the SMRP reservation path retransmission delay time. It presents the Path Retransmission by Crossover Router Scheme, Path Rerouting by Pointer Forwarding Scenario, and also the proposed Hybrid Resource Reservation Mechanism.
Chapter 5 presents the simulation of our research. It discusses the network topology and the parameters in the simulation study.
Chapter 6 describes the result of T-Test analysis and discusses the simulation results. It shows the effects of the algorithm on the performance of the network. The reservation handover delay, packet drop probability and throughput will be discussed in this chapter.
Chapter 7 presents the conclusion of this research. It summarizes the objective achievement, contributions and the findings of this dissertation, as well as the related work in future.