Joint activity and data detection for machine to machine communication via Bayes Risk optimization

Abstract

Abstract-Performing joint detection of activity and data is a promising approach to reduce management overhead in Machineto-Machine communication. However, erroneous activity detection has severe impacts on the system performance. Estimating an active node or user erroneously to be inactive results in a loss of data. To optimally balance activity and data detection, we derive a novel joint activity and data detector that bases on the minimization of the Bayes Risk. The Bayes Risk detector allows to control error rates with respect to the activity detection dynamically by a parameter that can be controlled by higher layers. In this paper we derive the Bayes Risk detector for a general linear system and present exemplary results for a specific Machine-to-Machine communication scenario.

Compressed sensing Bayes-risk detection for frame based multi-user systems

“…To this end, we proposed to augment joint activity and data detection by taking the risk for erroneous activity detection into regard resulting in the so called Bayes-Risk MAP detector introduced in [6], [7]. This detector takes cost for an erroneous activity detection into account which is done via the two factors As a major result of [6], the ratio of costs Ω = CFa CFi steers the detector between being more liberal and decide more in favor of activity or to be more conservative and decide more in favor of inactivity.…”

“…The reliability of the activity detection is crucial for the data-detection as erroneous activity detection strongly impacts the overall performance. In [6], [7] the authors have derived the Bayes-Risk detector for symbol-by-symbol joint activity and data detection. This detector allows to control the error rates with respect to activity detection by an additional risk parameter such that certain system dependent error rate requirements are met.…”