DMAC - An Adaptive Energy-Efficient and Low-Latency MAC for Data Gathering in Sensor Networks

In general, MAC protocols for sensor networks utilize activation/sleep duty cycles so as to improve energy efficiency and prolong network lifetime. However, this results in the data forwarding interruption problem, whereby nodes on a multihop path to the sink are not all awake when the data delivery is in progress. This results in sleep latency. The paper proposes DMAC, which eliminates the sleep latency by basing the active/sleep schedule of a node on its depth on the forwarding tree. Hence nodes wake up sequentially as data moves up the forwarding tree. The paper also proposes mechanisms to adapt the duty cycles based on the current traffic conditions in the network to further reduce sleep latency and improve energy efficiency. Moreover, they propose a data prediction mechanism and the use of more to send (MTS) packets to reduce the latency in scenarios where when a node has more data to send than it could possibly send in the current cycle, or when there is channel contention from its siblings. Though the approach works well for links which are pretty reliable, the delay in packet delivery increases significantly when the channel is noisy, since if a packet is lost during transmission, it can only be re-sent during the next cycle. The authors compare the performance of DMAC with SMAC and a MAC in which nodes are ON at all times. The results show that DMAC has lesser latency and is more energy efficient compared to SMAC.