Energy-Efficient Delay-Tolerant Communication: Revisiting Optimality of Superposition Coding in Broadcast Channels

Communication can consume a significant fraction of the available energy for many simple sensor devices for which battery life is an important consideration in deployment. Battery power is consumed not only by transmit power amplifier but also in the radio frequency circuits and digital processors during transmission and reception. When communication requirements are bursty, many devices incorporate a ‘sleep' state where the circuit power consumption is also reduced by turning circuits off. Delaying transmission can allow devices to sleep more and conserve energy. We consider the optimal tradeoff between receiver energy consumption and average throughput, and derive insights on multi-user downlink communication. We reformulate the problem with generalized power constraints on the transmitter's and the receiver's power consumption - depending on their states, either transmit/receive or sleep, they consume different amounts of power. We show how these changes of power constraints affect average spectral efficiency. In Gaussian broadcast channels, taking into account the receivers' power constraints, we show that multi-user transmission schemes, previously proven to be optimal for maximizing spectral efficiency, such as superposition coding and dirty paper coding (DPC) are not always optimal. We characterize the condition, under which these schemes remain optimal, in terms of receivers' power constraints. These models are suited for machine-to-machine (M2M) communications and wireless sensor networks where 1) transmitters and/or receivers are battery-powered devices, 2) their locations are static once deployed, and 3) their data characteristic is not delay-sensitive.