Multiuser Detection (MUD) Combined with array signal processing in current wireless communication environments 2002.4.10.Wed. 박사 3학기 구 정 회
Contents Introduction Motivation Definition System model Basic algorithms Space-time multiuser detection Current research issues Proposals Conclusion
Introduction Multiple access interference (MAI) Multiuser detection Interference between direct-sequence user A factor which limits the capacity and performance of DS-CDMA system As the number of interferers or their power increases, MAI becomes substantial Multiuser detection Area of research with potential to significantly improve DS-CDMA communication Combination with array signal processing
Motivation (1/2) Conventional detection Matched filtering + Sampling of the received signal + Decision device In a single path transmission environment Optimal in the sense that the SNR is maximized Maximum-likelihood (ML) detection In a multiuser environment The SNR is still maximized Not ML due to the presence of MAI
Motivation (2/2) CDMA is interference-limited system : NO ! Conventional detector is interference-limited : Sub-optimal approach Conventional detector does not take into account the existence of MAI Verdu showed that “ It is the thermal noise and not the MAI that rules the ultimate performance levels attainable in a CDMA system” Is it possible to exploit the particular structure of the MAI ? : YES ! Note Conventional detector (output of a bank of matched filter) provides a minimal sufficient statistics for detection Asynchronous CDMA system은 MAI외에, 채널에 의한 ISI가 생기는 dispersive CDMA system의 특별한 경우로 볼 수 있다.
Definition Code and timing information of multiple users are jointly used to better detect each individual user Important assumption The codes and timing information of the multiple users are known to the receiver a priori
Synchronous system model (1/2) Baseband received signal Output of the kth user’s correlator
Synchronous system model (2/2) Three user synchronous system : Matrix-vector system model
Asynchronous system model Baseband received signal
Basic algorithms Optimal multiuser detector Maximum likelihood sequence detector (’86, Verdu) Sub-optimal multiuser detectors Linear multiuser detectors Subtractive interference cancellation detectors Trade-off between complexity and performance
Optimal multiuser detector Solution to the ML problem Combinatorial quadratic minimization : NP-hard problem Only the exhaustive search will guarantee the global minimum
Linear multiuser detector Basic principle Apply a linear mapping, L, to the soft output of the conventional detector to reduce the MAI seen by each user Decorrelating detector MMSE detector
Decorrelating detector Applies the inverse of the correlation matrix Soft estimate of the detector All the MAI has been removed at the expense of noise enhancement Unconstrained, quadratic minimization problem
MMSE detector Take into account the background noise and utilizes knowledge of the received signal Linear mapping which minimizes the cost function Soft estimate of the MMSE detctor
Subtractive interference cancellation Basic principle The creation at the receiver of separate estimates of the MAI contributed by each user Successive interference cancellation (SIC) Parallel interference cancellation (PIC)
SIC Takes serial approach to canceling interference Implementation difficulties One additional bit delay is required per stage of cancellation There is a need to re-order the signals whenever the power profile changes
PIC Estimate and subtracts out all of the MAI for each user in parallel
Performance comparison of SIC and PIC Major disadvantage of nonlinear detectors Dependence on reliable estimates of the received amplitudes PIC SIC Advantages In a well power- controlled environments In fading environments Disadvantages Requires more hardware Problem of power reordering, Large delay
Performance analysis (1/2) (With perfect power control)
Performance analysis (2/2) (Flat Rayleigh fading with channel est.)
Space-time MUD (1/7) Problem of MUD in multipath CDMA channels with receiver antenna array
Space-time MUD (2/7) Signal model
Space-time MUD (3/7) Sufficient statistic Summarizes the useful information that a measurement brings about a parameter Find that for demodulating the multiuser symbols from the space-time signal Define the following,
Space-time MUD (4/7) Likelihood function of the received waveform conditioned on all the transmitted symbols of all users b Cameron-Martin formula
Space-time MUD (5/7) Sufficient statistic for detecting the multiuser symbol b : How to obtain ? Passing the received signal through (KL) beamformers directed to each path of each user’s signal, followed by a bank of K maximum-ratio multipath combiners (i.e. RAKE receiver) Sufficient statistic = Output of space-time matched filter Beamformer is a spatial matched filter RAKE receiver is a temporal matched filter
Space-time MUD (6/7) (Receiver structure)
Space-time MUD (7/7) Simulation environments 2 users, 2 multipath/user PG=128, 8 array elements array, SNR=-20dB
Current research issues (1/2) Choice for a practical MUD algorithm Complexity Performance Spatially and temporally noise-whitening receiver structure is developed
Current research issues (2/2) System design choices If MUD is to be part of the next standard, some minimum performance requirements have to be specified MUD research is still in in a phase that would not justify making it a mandatory feature for wideband CDMA standards Even though MUD is a receiver technique, it might have an impact on the system design because of its large complexity, while a proper system design might ease the implementation of the MUD
Proposals Subspace-based blind adaptive detector with lower computational complexity, robustness against signature waveform mismatch, non-Gaussian noise, impulsive noise Blind receiver for multiuser detection in unknown correlated noise Dual mode multiuser detector that dynamically switches its detection mode between matched-filter and decorrelator operations based on the channel characteristics
Conclusion Current wireless communication environments require considerable signal-processing ‘intelligence’ Two categories of many advanced signal processing techniques are multiuser detection and space-time processing Combined multiuser detection and array processing methods can ‘substantially’ outperform the conventional detector