Abstract

This thesis introduces new network coding techniques to improve the file sharing and video streaming performance of wireless star and mesh networks. In this thesis we propose a new XOR based scheduling algorithm for network coding in cooperative local repair. The proposed algorithm commences in three phases. In the first phase, nodes exchange packets availability vectors. These vectors are functions of the probability of correct packet reception over the channel. This is followed by a short period of distributed scheduling where the nodes execute the processing algorithm which tries to minimize the total transmission time. In the third phase, nodes transmit the encoded packets as per the decision of the scheduling algorithm. Simulation results show improvement in system throughput and processing delay for the proposed algorithm. We also study the trade-offs between file sizes, processing delays, number of users and packet availability. In the sequel we display the favorable effects of file segmentation on the performance of the proposed scheduling algorithm. Furthermore, the upper bound on the performance and the analysis of the proposed scheduling algorithm are derived.
Also, in this thesis, the effects of random network coding on code division multiple access/time division duplex (CDMA/TDD) platforms for wireless mesh networks are studied and evaluated. A multi-hop mesh network with single source and multiple receiving nodes is assumed. For reliable data transfer, a Selective Repeat ARQ protocol is used. Two scenarios are evaluated for their efficiency. In scenario 1, but not in scenario 2, random network coding is applied to CDMA/TDD wireless mesh networks. The delay and delay jitter for both scenarios are computed. The study also focuses on the effects of uncontrolled parameters such as the minimum number of neighbors and the network connectivity, and of controlled parameters such as Galois Field (GF) size, packet size, number of Walsh functions employed at each node and the Processing Gain. The analysis and simulation results show that applying random network coding to CDMA/TDD systems in wireless mesh networks could provide a noticeable improvement in overall efficiency.
We also propose a cross layer approach for the Random Network coded-Code Division Multiple Access/Time Division Duplex (RNC-CDMA/TDD) wireless mesh networks. The proposed algorithm selects the number of assigned Walsh functions depending on the network topology. Two strategies of Walsh function assignments are proposed. In the first, nodes determine the number of their assigned Walsh functions depending on the neighbor with the maximum number of neighbors, which we call the worst case assignment. In the second, nodes determine the number of their assigned Walsh functions depending on the need for each transmission. Simulation results show the possible achievable improvement in the system performance, delay and delay jitter due to cross layer design.