Optimal Age-Energy Trade-off via Sleep-Wake Scheduling

The problem of controlling and analyzing information updates has received significant recent attention. In particular, the past few years have seen a dramatic increase in the design of age-based multi-source scheduling policies. However, many information generating nodes are often battery operated where the lifetime of the network is a critical element. Thus, an important open question is how to minimize the age of information (AoI) in such energy constrained networks? In this work, we address this question by leveraging carrier sensing in order to allow information sources to sleep when they are not transmitting, thereby saving precious battery power, and elongating the lifetime of the sources. Thus, a source sleeps if it senses that the channel is busy, or upon completing a packet transmission. The introduction of sleep periods, however, increases the AoI. Thus, we design sleeping periods to minimize the total weighted average peak age, while simultaneously meeting a per-source energy constraint. The problem of minimizing the cumulative age of sources under energy constraints is a hard non-convex optimization problem. However, when the carrier sensing time is negligible compared to the packet transmission time, we are able to provide a near-optimal solution. Moreover, we represent our solution in a simple form that allows us to design a simple-to-implement algorithm to obtain its value. Finally, we show that the performance of our proposed algorithm is asymptotically no worse than the optimal synchronized scheduler.