Optimum multiuser detection (MUD) with an unknown number of active users requires simultaneous estimation of the active-user set, their unknown parameters and their transmitted data. Recent advances in MUD have shown that optimum receivers for this system model can be obtained using random-set theory (RST), a widely used tool in radar signal processing. Despite the capability of generalizing standard MUD to the case of unknown number of users, RST offers a way to perform detection of user log in and out by means of Bayesian recursions (BR). While previous works have concentrated on the definition of optimal detectors, design of efficient receivers was not addressed. Indeed, implementation of optimum detectors may be limited by their complexity, which grows exponentially with the number of potential users. The aim of this paper is to show that this computational burden can be drastically reduced, with little or no loss of performance, by applying a suitable version of the sphere detection (SD) algorithm. If users’ continuous parameters are known, SD algorithm allows exact implementation of the optimal detector under one-shot scenarios at polynomial complexity for moderate signal-to-noise ratio (SNR), while requiring a suitable approximation of BRs in a dynamic environment. The approach is also extended for cases wherein continuous parameters are unknown. The developed detectors are compared against their optimal counterparts and their effectiveness is shown through numerical simulations
Low-complexity receivers for multiuser detection with an unknown number of active users
LOPS, Marco
2010-01-01
Abstract
Optimum multiuser detection (MUD) with an unknown number of active users requires simultaneous estimation of the active-user set, their unknown parameters and their transmitted data. Recent advances in MUD have shown that optimum receivers for this system model can be obtained using random-set theory (RST), a widely used tool in radar signal processing. Despite the capability of generalizing standard MUD to the case of unknown number of users, RST offers a way to perform detection of user log in and out by means of Bayesian recursions (BR). While previous works have concentrated on the definition of optimal detectors, design of efficient receivers was not addressed. Indeed, implementation of optimum detectors may be limited by their complexity, which grows exponentially with the number of potential users. The aim of this paper is to show that this computational burden can be drastically reduced, with little or no loss of performance, by applying a suitable version of the sphere detection (SD) algorithm. If users’ continuous parameters are known, SD algorithm allows exact implementation of the optimal detector under one-shot scenarios at polynomial complexity for moderate signal-to-noise ratio (SNR), while requiring a suitable approximation of BRs in a dynamic environment. The approach is also extended for cases wherein continuous parameters are unknown. The developed detectors are compared against their optimal counterparts and their effectiveness is shown through numerical simulationsI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
