Design and optimization of multiple access protocols for ad hoc wireless networks

The main objective of medium access control (MAC) protocols for wireless networks is to share efficiently and fairly communication medium among many contending users. In wireless ad hoc networks, multiple stations use the radio channel to communicate without presence of any fixed infrastructure. In this case, multiple access is basically distributed and random, and collisions are unavoidable. In this thesis, we address design and optimization issues of random MAC protocols for wireless ad hoc networks. We have focused on two fundamental problems that affect their performances: The backoff scheme design and optimization, and the robust handling of the hidden terminal problem. The first problem is inherent to any random access mechanism as collisions of packets are unavoidable. Thus, the performances of the MAC protocol in this case are mainly governed by the efficiency of its retransmission mechanism. The second problem is associated to the multihop situations faced in ad hoc networks. It is shown that in this case, known efficient access protocol in single hop network, as the IEEE 802.11 DCF, suffers serious performance degradation due to the lack of unique broadcasting channel to support efficient signaling schemes. In the first part of the thesis, we address the optimization of the retransmission schemes for ALOHA and DCF protocols in single hop networks. Optimization is carried out in order to maximize the overall network throughout. Our goal was then to design simple and blind optimal retransmission mechanisms under this configuration. This was achieved more efficiently for DCF by exploiting the asymmetry of idle and collision events' duration in carrier sense systems. Multihop networks where addressed in the second part of the thesis. For DCF protocol, we propose a new handshaking mechanism, RTS/Repeated-CTS, that handle effectively the hidden node problem, but at the expense of increased signaling overhead. The Backoff scheme optimization was then addressed by arguing the use of a stochastic approximation (SA) algorithm to compensate the lack of symmetric channel behavior at the users. In the perspective of using ultra-wideband (UWB) signaling technology for ad hoc networks, and to take benefit from its multiplexing capabilities, we have introduced a new multi-channel ALOHA-like protocol. The proposed protocol does not use carrier sensing due its inefficiency in multihop topologies, so the protocol is insensitive to the hidden node problem. We show further that the proposed protocol assimilates the behavior of CSMA protocol in a multi-channel multihop system. We then addressed fundamental performance of time-delay estimation schemes in impulse radio UWB systems. This issue is crucial for the deployment of ad hoc based UWB solutions as the synchronization task of sub nanosecond signals appears to be a critical issue. Fundamental time-delay estimation performances were derived in term of lower and upper bounds on the mean square error (MSE) of different estimators. The bounds we have used in this work were both derived by us. The lower bound is an improvement of the Ziv-Zakai Lower bound, and it is the tightest known lower bound on MSE of any estimator. In the spirit of the ZZLB, we propose also an upper bound that applies for any estimator, and that is simple to derive with controllable accuracy. The bounds were then applied to characterize performance of perfect maximum likelihood (ML) estimator, mis-matched ML estimator, equal gain combining estimator, and energy detector estimator. Different results on estimators' performances were provided that have shown that the simple energy detector performs well with few numbers of signal's repetition and adequate integration window length.