DISTRIBUTED SURVIVABLE PIPELINE COMPUTATION AND COMMUNICATION

The increasing availability of mobile and/or intelligent sensors with computation and wireless communication capabilities makes computer networks more and more heterogeneous and complex. At the one end of these networks are the grid networks for high performance com-ponents with fixed network structures and fixed roles of components. On the other end of the networks for the low power and/or low per-formance components there are the mesh networks with adaptive tree structures in which data is sent to the root for processing. The combi-nation of the above structures would create a heterogenous solution for heterogenous networks. This solution would create an adaptive graph structure among the components to process data within the network therefore reducing the communication need of tree structure. The components have adaptive roles to maintain the tasks of the network when components brakes down or reappears, and also, when latencies and bandwidths are varying within the network. In this dissertation I propose a novel protocol to satisfy these requirements.