Abstract

Mobile Ad Hoc Networks (MANETs) are gaining increasing popularity in recent years because of their ease of deployment. They are distributed, dynamic, and self-configurable without infrastructure support. Routing in ad hoc networks is a challenging task because of the MANET dynamic nature. Hence, researchers were focused in designing best-effort distributed and dynamic routing protocols to ensure optimum network operations in an unpredictable wireless environment. Nowadays, there is an increased demand on multimedia applications (stringent delay and reliability requirements), which makes a shift from best-effort services to Quality of Services.

Actually, the challenge in wireless ad hoc networks is that neighbor nodes share the same channel and they take part in forwarding packets. Therefore, the total effective channel capacity is not only limited by the raw channel capacity but is also limited by the interactions and interferences among neighboring nodes. Thus, such factors should be taken in consideration in order to offer QoS routing. While, some of the distributed QoS route selection algorithms assume the availability of such information, others propose mechanisms to estimate them.

The goals of this thesis are: (i) to analyze the performance of IEEE 802.11 MAC mechanism in non-saturation conditions, (ii) to use the analysis in the context of multi-hop ad hoc networks, (iii) to derive theoretical limits for nodes performance in multi-hop ad hoc networks, (iv) to use the multi-hop analysis in QoS route selection.

We start the thesis by proposing a discrete-time 3D Markov chain model to analyze the saturation performance of the RTS/CTS access mode. This model integrates the backoff countdown process, retransmission retry limits, and transmission errors into one model. The impact of system parameters (e.g., number of nodes, packet size, retry limits, and BERs) are analyzed. Next, we extend the 3D model to analyze the performance under non-saturation conditions and finite buffer capacity using two different approaches. First, we extend the 3D model into a 4D model to integrate the transmission buffer behavior. Second, we replace the 4D model by an M/G/1/K queueing system model with independent samples from the saturation analysis. The latter model gives similar results as the former but with a reduction in the analysis complexity. Next and by means of the non-saturation analysis, we proposed an approximate mathematical model for multi-hop ad hoc networks. Furthermore, we proposed an iterative mechanism to estimate the throughput in the presence of multiple flows. Finally, we used the multi-hop analysis to propose a QoS route selection algorithm. In this algorithm, we concentrate on the throughput as a QoS parameter. However, the proposed algorithm is valid to be used with other QoS parameters, such as packet delay, packet loss probability, and fairness. Analytical and simulation results show the deficiency of the current route selection algorithm in AODV and at the same time verifies the need for QoS route selection algorithms.