1:00, SPCOM-L7.1
A BLIND NETWORK OF EXTENDED KALMAN FILTERS FOR NONSTATIONARY CHANNEL EQUALIZATION
R. AMARA, S. MARCOS
In this paper, a blind Network of Extended Kalman Filters (NEKF) is introduced for nonstationary linear channel equalization. The structure of NKF was recently suggested for optimal channel equalization. As the knowledge of the channel is the main constraint within the NKF-equalizer, we here propose to extend the state to estimate, that was previously formed by the last M transmitted symbols, to the time-varying channel coefficients. The observation model becomes nonlinear suggesting thus extended Kalman filtering for state estimation. The proposed NEKF algorithm is completely blind towards any learning phase, with fast convergence properties. Compared to the blind bayesian estimator proposed by Iltis et al., the NEKF-based equalizer shows good performance with a really lower complexity.

1:20, SPCOM-L7.2
BLIND CHANNEL IDENTIFICATION BY SUBSPACE TRACKING AND SUCCESSIVE CANCELLATION
X. LI, H. FAN
Traditional subspace method (SS) for blind channel identification require accurate rank estimation with a computational complexity of O(m^3), where m is the data vector length. In this paper, we introduce new adaptive subspace algorithms using ULV updating and successive cancellation techniques. In addition to reduce the computational complexity to O(m^2), the new algorithms do not need to estimate the subspace rank. Channel length can be over-estimated during the subspace tracking and channel vector optimization steps. It can then be recovered at the end by a successive cancellation procedure. Simulation shows that the new algorithms outperform the traditional SS methods in case the subspace rank is difficult to estimate.

1:40, SPCOM-L7.3
LINEAR PREDICTION ERROR METHOD FOR BLIND IDENTIFICATION OF TIME-VARYING CHANNELS: THEORETICAL RESULTS
J. TUGNAIT
Blind channel estimation for SIMO time-varying channels is considered using only the second-order statistics of the data. The time-varying channel is assumed to be described by a complex exponential basis expansion model (CE-BEM). The linear prediction error method for blind identification of time-invariant channels is extended to time-varying channels represented by a CE-BEM. Sufficient conditions for identifiability are investigated. Cyclostationary nature of the received signal is exploited to consistently estimate the time-varying correlation function of the data from a single observation record. The focus of the paper is on certain theoretical issues.

2:00, SPCOM-L7.4
DIRECT AND EM-BASED MAP SEQUENCE ESTIMATION WITH UNKNOWN TIME-VARYING CHANNELS
H. CHEN, R. PERRY, K. BUCKLEY
In this paper we address sequence estimation when the InterSymbol Interference (ISI) communication channel is unknown and time varying. We employ a Maximum A Posterior (MAP) approach, in which the unknown channel parameters are assigned a distribution and integrated out. For several channel models of interest we describe both the exact MAP estimator and Viterbi algorithm based implementations. We also present EM algorithms for solving these MAP sequence estimation problems, and we contrast these EM solutions with direct MAP algorithms. http://www.ece.villanova.edu/~perry/

2:20, SPCOM-L7.5
ON THE EQUALIZABILITY FOR NONLINEAR/TIME-VARYING MULTI-USER CHANNELS
S. EL ASMI, M. MBOUP
We attack the problem of perfect equalizability of multi-user channels, in which the usual linear time-invariant assumption is dismissed. In the linear, time-invariant case, condition for perfect equalizability is plain andexpressed in terms of the column rank of the channel's transfer matrix. Using the module-theoretic approach developped by Fliess, in which the transfer matrix of a time-varying channel as well as the rank of a non-linear channel are clearly defined, we show how the condition obtained in the linear time-invariant case naturally extends to the time-varying and the non-linear cases.

2:40, SPCOM-L7.6
TRANSMITTER CHANNEL TRACKING FOR OPTIMAL POWER ALLOCATION
F. REY, M. LAMARCA, V. GREGORI
The design of accurate equalization schemes, optimizing the transmission strategy, can be achieved if channel status information is available not only at the receiver but also at the transmitter side. Accordingly, we propose a feasible scheme to track the channel response by the transmitter based on channel prediction, becoming a suitable solution in time-varying channels. Moreover, to follow faithfully the channel variability, the receiver can estimate the channel prediction error and next update the transmitter predictor through a return link. This paper, provides an accurate study of the predictor design, and analyzes the way to minimize the amount of information exchanged through the feedback channel concerning the differential entropy of channel evolution. Furthermore, the prediction error is quantified focusing on the rate-distortion function, and it is shown that a low throughput is enough for tracking the channel coefficients even in the presence of fast time-varying channels