Delay and capacity in energy efficient sensor networks

MAC protocols for wireless sensor networks employ periodic switching to low energy sleep state in order to enhance network lifetime. During the sleep state, the sensors do not perform energy consuming operations such as receiving and transmitting packets. During the normal state, CSMA based multi-access mechanism is the MAC protocol of choice in distributed, unsynchronized sensor networks. The energy conserving mechanism has a two-fold effect on delay in the network. On one hand it increases delay since many a times the intended receiver may be in sleep state and the transmitter has to delay the transmission to allow the receiver to wake up. On the other hand, since the sensors do not transmit in sleep state, the contention for channel is reduced which tends to improve delay. In this paper we present a queuing theoretic analysis of delay and capacity in sensor networks with uncoordinated sleep mechanism and characterize the energy-delay-capacity tradeoffs. We consider several sleep states which consume different levels of energy. We model sensor networks as queuing networks and evaluate closed form expressions for average packet delay and maximum achievable per-node throughput in terms of network parameters and sleep schedule. Comparisons with the performance of networks that do not employ any energy conserving mechanisms show that any of the energy conserving sleep states in the networks considered in this paper leads to considerable degradation in delay and capacity of the network.