Advanced receivers for high speed downlink packet access in UMTS

The thesis presentation covers some of the advanced signal processing techniques we proposed for Universal Mobile Telecommunications System (UMTS) user equipment receivers, which are particularly suited for the so-called 3.5G High Speed Downlink Packet Access (HSDPA) services. First we give a brief description of the UMTS downlink system and the HSDPA specifications. In the second part we consider a class of user dedicated downlink channel estimation methods, which are especially beneficial when there is dedicated channel transmit beamforming. The methods do not assume any a priori knowledge of the path delays and the beamforming parameters and they exploit all the transmitted pilot sequences as well as the structured dynamics of the channel. First we build least squares (LS) estimates of the channels associated with dedicated and common pilots in every slot by the aid of available pilot sequences. Then we optimally improve the dedicated channel estimate quality by jointly Kalman filtering the two LS estimates or alternatively (sub-optimally) Kalman filtering them separately and combining via weighted LS. In the suboptimal case, the order of Kalman filtering and weighted LS combining results in differing performance and complexity in different conditions. In order to estimate the missing model parameters we incorporate the expectation maximization (EM) algorithm to the Kalman filtering mechanism by expanding the latter by one-lag smoothing. In the third part we assess the benefits of using chip level equalizers w.r.t. the usage of the conventional Rake receiver and using hard decision or hyperbolic tangent symbol nonlinearities w.r.t. the usage of linear feedback operations in the context of an iterative parallel interference cancellation receiver that we derive from the polynomial expansion of the symbol level covariance matrix inverse after the first stage equalization. Since the equalizers at different stages of the considered interference canceller are to be different as well, in order to estimate the essential equalizer parameters of a particular stage we use the analysis results from the preceding stages. We conclude by listing the open points and the possible future work.