Co-sponsored by: Anader Benyamin-Seeyar

This talk considers a massive device communications scenario in which a large number of devices need to connect to a base-station, but user traffic is sporadic so that at any given coherence time only a subset of users are active. For such a system, user activity detection and channel estimation are key issues. This talk first provides an information theoretical analysis for massive connectivity by illustrating how the cost of user identification and channel estimation affects the achievable degree-of-freedom. Next, we present a two-phase framework in which compressed sensing techniques are used in the first phase to identify the devices and their channels, while data transmission takes place in the second phase. We propose the use of approximate message passing (AMP) for device identification and show that state evolution can be used to analytically characterize the missed detection and false alarm probabilities in AMP. This talk further considers the massive connectivity problem in the massive MIMO regime. We analytically show that massive MIMO can significantly enhance user activity detection, but the non-orthogonality of pilot sequences can nevertheless introduce significant channel estimation error, hence limiting the overall rate.

We quantify this effect and characterize the optimal pilot length for massive uncoordinated device access.

Speaker(s): Dr. Wei Yu,

Location:
Montreal, Quebec