Position-based routing and MAC protocols for wireless ad-hoc networks

Abstract

This thesis presents the Forecasting Routing Technique (FORTEL), a routing protocol for Mobile Ad-Hoc Networks (MANETs) based on the nodes’ Location Information. FORTEL stores the nodes’ location information in the Location Table (LT) in order to construct routes between the source and the destination nodes. FORTEL follows the source routing strategy, which has rarely been applied in position-based routing. According to the source routing strategy, the end-to-end route is attached to the packet, therefore, the processing cost, in regards to the intermediate nodes that simply relay the packet according to route, is minimized. FORTEL’s key mechanisms include: first, the location update scheme, employed to keep the LT entries up-to-date with the network topology. Besides the mobility variation and the constant rate location update schemes applied, a window location update scheme is presented to increase the LT’s information accuracy. Second, the switching mechanism, between “Hello” message and location update employed, to reduce the protocol’s routing overhead. Third and most important is the route computation mechanism, which is integrated with a topology forecasting technique to construct up-to-date routes between the communication peers, aiming to achieve high delivery rate and increase the protocol robustness against the nodes’ movement. FORTEL demonstrates higher performance as compared to other MANET’s routing protocols, and it delivers up to 20% more packets than AODV and up to 60 % more than DSR and OLSR, while maintaining low levels of routing overhead and network delay at the same time. The effectiveness of the window update scheme is also discussed, and it proves to increase FORTEL’s delivery rate by up to 30% as compared to the other update schemes. A common and frequently occurring phenomenon, in wireless networks, is the Hidden Terminal problem that significantly impacts the communication performance and the efficiency of the routing and MAC protocols. Beaconless routing approach in MANETs, which delivers data packets without prior knowledge of any sort `of information, suffers from packet duplication caused by the hidden nodes during the contention process. Moreover, the throughput of the IEEE MAC protocol decreases dramatically when the hidden terminal problem occurs. RTS/CTS mechanism fails to eliminate the problem and can further degrade the network’s performance by introducing additional overhead. To tackle these challenges, this thesis presents two techniques, the Sender Suppression Algorithm and the Location-Aided MAC, where both rely on the nodes’ position to eliminate packet duplication in the beaconless routing and improve the performance of the 802.11 MAC respectively. Both schemes are based on the concept of grouping the nodes into zones and assign different time delay to each one. According to the Sender Suppression Algorithm, the sender’s forwarding area is divided into three zones, therefore, the local timer, set to define the time that the receiver has to wait before responding to the sender’s transmission, is added to the assigned zone delay. Following the first response, the sender interferes and suppresses the receivers with active timer of. On the other hand, the Location-Aided MAC, essentially a hybrid MAC, combines the concepts of time division and carrier sensing. The radio range of the wireless receiver is partitioned into four zones with different zone delays assigned to each zone. Channel access within the zone is purely controlled by CSMA/CA protocol, while it is time-based amongst zones. The effectiveness of the proposed techniques is demonstrated through simulation tests. Location-Aided MAC considerably improves the network’s throughput compared to CSMA/CA and RTS/CTS. However, remarkable results come when the proposed technique and the RTS/CTS are combined, which achieves up to 20% more throughput as compared to the standalone RTS/CTS. Finally, the thesis presents a novel link lifetime estimation method for greedy forwarding to compute the link duration between two nodes. Based on a newly introduced Stability-Aware Greedy (SAG) scheme, the proposed method incorporates the destination node in the computation process and thus has a significant advantage over the conventional method, which only considers the information of the nodes composing the link.