A Survey on Relay Selection for Geographical Forwarding in Sleep-Wake Cycling Wireless Sensor Networks

Abstract

Our work is motivated by geographical sending of sporadic alert packets to a base station in a wireless sensor network (WSN), where the nodes are sleep-wake cycling periodically and nonconcurrently. We seek to develop neighborhood sending calculations that can be tuned in order to tradeoff the end-to-end delays against an aggregate cost, for example, the jump tally or aggregate energy. Our approach is to solve, at each sending node enroute to the sink, the neighborhood sending problem of limiting one-jump holding up delay subject to a lower bound limitation on a suitable reward offered by the next-bounce relay; the requirement serves to tune the tradeoff. The reward metric used for the nearby problem is based on the end-to-end add up to cost objective (for instance, when the aggregate cost is bounce check, we choose to use the progress toward sink made by a relay as the reward). The sending node, in any case, is uncertain about the number of relays, their wake-up times, and the reward values, however knows the likelihood appropriations of these quantities. At each relay wake-up moment, when a relay reveals its reward value, the sending node's problem is to forward the packet or to sit tight for further relays to wake-up. In terms of the operations research literature, our work can be considered as a variation of the asset selling problem. We formulate our nearby sending problem as an incompletely observable Markov decision process (POMDP) and acquire inner and outer limits for the ideal arrangement. Motivated by the computational complexity involved in the policies derived out of these limits, we formulate an alternate simplified model, the ideal strategy for which is a simple threshold rule. We provide reproduction results to compare the performance of the inner and outer bound policies against the simple strategy, and furthermore against the ideal approach when the source knows the exact number of relays. Observing the great performance and the ease of implementation of the simple approach, we apply it to our persuading problem, i.e., neighborhood geographical steering of sporadic caution packets in a large WSN. We compare the end-to-end performance (i.e., average aggregate delay and average aggregate cost) obtained by the simple strategy, when used for neighborhood geographical sending, against that obtained by the all inclusive ideal sending calculation proposed by Kim et al.