On bounding node-to-sink latency in wireless sensor networks with multiple sinks

Bounding node-to-sink latency is an important issue of wireless sensor networks (WSNs) with a quality of service requirement. This paper proposes to deploy multiple sinks to control the worst case node-to-sink data latency in WSNs. The end-to-end latency in multihop wireless networks is known to be proportional to the hop length of the routing path that the message moves over. Therefore, we formulate the question of what is the minimum number of sinks and their locations to bound the latency as the minimum d-hop sink placement problem. We also consider its capacitated version. We show problems are NP-hard in unit disk graph (UDG) and unit ball graph, and propose constant factor approximations of the problems in both graph models. We further extend our algorithms so that they can work well in more realistic quasi UDG model. A simulation study is also conducted to see the average performance of our algorithms.