Exploitation of path diversity in cooperative multi-hop wireless networks

Radio-equipped mobile computers can form a mobile ad hoc multi-hop wireless network. When these mobile nodes are highly dynamic, the task to find a route from one node to another is very challenging. In order to design an efficient and effective routing system, all the characteristics of mobile wireless communication should be exploited. Clearly, one of the promising techniques is the exploitation of the path diversity. At a given time, channel condition in any link varies as nodes move; often the variation is significant even when the mobility is moderate. As a result, the best path between two nodes will also vary. Consequently, links and paths will vary in space and time. This variability is called diversity. The main goal in this thesis is to exploit the path diversity in order to improve end-to-end performance metrics such as throughput and delivery probability, but without greatly increasing overhead.;The first step in this research is to develop several path quality metrics, which we will seek to optimize by exploiting path diversity. For example, path quality is the smallest SNR over all links along the path. In general, path quality will be defined based on the protocol designer's routing objective (e.g., maximizing the throughput). Once possible metrics have been defined, we explore an idealized and aggressive path diversity exploitation technique to determine the upper limit of the benefits that can be achieved by exploiting path diversity. We also explore the path differences resulting from different path metrics.;Our idealized and aggressive approach to diversity exploitation results in too much overhead to be of practical use. Thus, we focus on developing efficient path diversity exploitation techniques. To this end, the qualities of the paths are monitored reactively; when the current best path drops below a threshold, a local search to exploit path diversity is triggered. To further reduce the overhead and find the new best path efficiently, two more methods, namely, J-test and routing metric based power control, are proposed. Additionally, a novel routing technique for automatic stretching and shrinking on the current best path is proposed for dynamic route adjustments.;Another goal in this thesis is to utilize the uncertainty of packet transmissions to the intended nodes that are prioritized by some criteria by grafting the path diversity exploitation onto opportunistic forwarding scheme. First of all, a method to construct the intended prioritized nodes based on paths' qualities is proposed. For the purpose of comparison between deterministic forwarding (resulting in a best path) and opportunistic forwarding (resulting in an opportunistic path), three protocols are proposed; one is for deterministic forwarding, another for pure opportunistic forwarding, and the other for opportunistic forwarding with some features used in deterministic forwarding. The level of opportunism depends on the relationship between packet error probability and SNR. The less steep the relationship is (i.e., the smaller the first derivative), the higher uncertainty of packet transmissions, this is, the better performance of an opportunistic approach. The comparison is performed with six different curves of the relationship; one is directly based on the standard radio model and the others are artificially derived based on the standard one.;A final part of this research focuses on developing techniques to track the relationship between packet error probability (PEP) and SNR. The various representative PEP/SNR relationships are determined from packet-level simulation in advance and then a relationship among them is estimated from the observations measured in real networks. The sequence of the estimated relationships over time provides useful information about the prediction of the future PEP/SNR relationship. This present and future channel estimation will help a cognitive routing protocol to achieve its intelligent task.