Dynamic Huffman addressing in wireless sensor networks

This paper introduces a new dynamic addressing technique for prolonging the operational duration of duty cycled wake-up radios found in wireless sensor networks. A wireless sensor network (WSN) consists of a field of spatially distributed wireless nodes which operate on minimal battery power for extremely long times. A WSN typically uses a multi-hop architecture where messages containing sensed data at the point of origin move through the network eventually finding a gateway node. To minimize energy consumption nodes in a WSN employ novel techniques such as powering down as many components as possible as often as possible. One such MAC technique used is to power down the entire radio, waking up to listen for data addressed to the node at a well known time. This listen time is referred to as the check period. If the nodes can maintain sufficient time synchronization the time between check periods can be extended, reducing the duty cycle and increasing the operational life of the WSN at the expense of data latency. Traditionally, the size of the addresses used throughout such a WSN is static and uniform.

Our proposed system is neither. As is typical in the WSN space we take a systems level perspective and seek to optimize global objectives. Specifically we consider a WSN field of nodes and maximize the useful life of the network. We assume that when any node has insufficient power to be useful, events can no longer be propagated through the WSN and the WSN is no longer operative. Our approach lengthens the life of the WSN by being especially conservative with nodes having low battery power. We investigate prefix-free (Huffman) encoding of the node addresses based on the energy map. That is, nodes with little battery life left will have the advantage of a short address, and therefore check period, so they are able to power down more quickly, while nodes with more battery life left will get longer addresses and require a longer check period. We will review techniques wi- – th a similar objective, offer a formal description of our algorithm and demonstrate that this approach prolongs the usefulness of certain sensor networks via Matlab simulation. Our simulation results show that our dynamic addressing technique can lengthen the operational life of a WSN by more than 10% over traditional fixed length addressing.