Compressive sensing based multi-user detection for machine-to-machine communication

ABSTRACTWith the expected growth of machine‐to‐machine communication, new requirements for future communication systems have to be considered. More specifically, the sporadic nature of machine‐to‐machine communication, low data rates, small packets and a large number of nodes necessitate low overhead communication schemes that do not require extended control signaling for resource allocation and management. Assuming a star topology with a central aggregation node that processes all sensor information, one possibility to reduce control signaling is the estimation of sensor node activity.In this paper, we discuss the application of greedy algorithms from the field of compressive sensing in a channel coded code division multiple access context to facilitate a joint detection of sensor node activity and transmitted data. To this end, a short introduction to compressive sensing theory and algorithms will be given. The main focus, however, will be on implications of this new approach. Especially, we consider the activity detection, which strongly determines the performance of the overall system. We show that the performance on a system level is dominated by the missed detection rate in comparison with the false alarm rate. Furthermore, we will discuss the incorporation of activity‐aware channel coding into this setup to extend the physical layer detection capabilities to code‐aided joint detection of data and activity. Copyright © 2013 John Wiley & Sons, Ltd. Future M2M systems require novel communication schemes with strongly reduced signaling overhead. To this end, we investigate CS algorithms for physical layer detection to jointly detect the activity and data of sensor nodes. We show that the activity detection performance of CS detectors is dominated by the missed detection rate in contrast to the false alarm rate, which motivates future development of novel algorithms, and furthermore, we incorporate activity‐aware channel coding to enable a code‐aided joint detection of data and activity.