Study of the impact of sleep phase duration on energy consumption in sensor chain

Wireless sensor networks (WSN) is a fast developing technology used in area of monitoring processes and objects. WSN differs significantly from traditional computer networks. There is no regular administration, radio interference should be taken into account, etc. Another feature of WSNs is limitation of power consumption due to the fact that the nodes are powered by batteries. The presence of such restrictions forces development of new routing protocols. There is a number of routing protocols built on the principles of energy efficiency. In present paper we consider PEGASIS protocol. This protocol utilizes idea of logical chain. Each node communicates with its nearest neighbours only. The work cycle of WSN is divided into rounds. On each round network randomly chooses a sink-designated node which will accumulate information. Sink notifies other nodes about its location sending a service message. This message travels along the chain in multihop manner, from node to node. When the last node receives such a message it replies with a message containg measured data. Data message moves towards the sink node. Each node adds its data to the message and passes it to the neighbour. Round ends when sink have recieved information from all the nodes. In present paper we investigate impact of adding sleep phase in working cycle of nodes. Sleep phase is an important energy saving measure broadly used in real-world sensor networks. But in case of a sensor chain each sleeping node will slow down the movement of messages which implies deceleration of data acquisiotion. There is a conjecture that this slowdown will be compensated by more economical energy spending. We study this issue using a simulation model of a sensor chain. Routing is based on PEGASIS protocol. Each sensor can be in one of two phases - active phase or sleep phase. Duration of each phase is a random value distributed by geometrical law (which responds to Chiasserini-Garetto model). Our simulation results show that in case of random moments of node falling asleep and waking up energy spending under some circumstances may exceed several times compared to the case of no sleep phase. One of main reasons for this situation is idle listening. Our results demonstrate the need to coordinate sleep schedules between WSN nodes.

Keywords

Authors

References