Synchronization algorithms for FBMC systems

Recently, filter bank multicarrier (FBMC) systems have received considerable attention for wired and wireless high-data-rate transmissions in frequency selective channels. Conventional multicarrier systems, known as orthogonal frequency division multiplexing (OFDM) systems, provide robustness to multipath channels, thanks to the introduction of a cyclic prefix (CP) that efficiently combats the intersymbol interference (ISI) in dispersive channels. However, the insertion of CP is pure redundancy, that decreases the spectral efficiency. Moreover, in OFDM systems the adopted pulse-shaping filter is a rectangular function, which exhibits poor frequency-decay. On the contrary, FBMC systems employ band limited pulse-shaping filters that overlap in time. This involves several advantages such as reduced sensitivity to narrowband interference, high flexibility to allocate group of subchannels to different users and a high spectral containment. The computational complexity of FBMC systems is higher than that of CP-OFDM systems. However, since the subchannel filters are obtained by complex modulation of a single filter, efficient polyphase implementations are possible. FMBC systems referred to as Filtered Multitone (FMT) systems have been proposed for high-speed digital subscriber line (VDSL) standards and are under investigation also for broadband wireless applications. FBMC systems based on offset quadrature amplitude modulation (OQAM), known as OFDM/OQAM systems, have been considered by the 3GPP standardization forum for improved downlink UTRAN interfaces. As all the multicarrier modulation schemes, one of the major disadvantages of FBMC systems is their sensitivity to carrier frequency and symbol timing errors. Specifically, phase noise and misalignments in time and frequency can considerably degrade the performance of FMT and OFDM/OQAM systems, giving rise to interference between successive symbols and adjacent subcarriers. Therefore, reliable and accurate symbol timing and carrier-frequency offset (CFO) synchronization schemes must be implemented for these systems. Several studies have been focused on parameter estimation for FBMC systems based on data-aided or blind techniques. In the first case it is in demand the transmission of known sequences or the use of a training symbol with a known structure while blind estimation algorithms use exclusively the statistic properties of the transmitted signal. In this thesis the problem of CFO and symbol timing synchronization is examined and new data-aided and blind estimation techniques are proposed. Specifically, it is presented a new joint symbol timing and CFO synchronization algorithm based on the least squares (LS) approach, which exploits the known structure of a training sequence made up of identical parts. This method, as illustrated by numerical simulations, can assure in a multipath channel sufficiently accurate symbol timing and CFO estimates. Moreover, the joint ML phase offset, CFO and symbol timing estimator for a multiple access (MA) OFDM/OQAM system is considered. The derived estimator exploits a short known preamble embedded in the burst of each of U users. Under the assumption that the CFO of each user is sufficiently small, the considered approach leads to U different approximate ML (AML) joint phase offset, CFO and symbol timing estimators. In particular, the phase and CFO estimators are in closed form, while the AML symbol timing estimator requires a one-dimensional maximization procedure. As regards blind synchronization techniques, it is proposed a closed-form CFO estimator based on the best linear unbiased (BLU) estimation principle for FMT systems. Although the BLU estimator is derived under the hypothesis of additive white Gaussian noise (AWGN) channel, it demonstrates a remarkable robustness against multipath fading. Blind CFO estimators based on the ML principle and obtained under the hypothesis of low SNR are also considered. Specifically, the proposed CFO estimators can exploit both the conjugate and the unconjugate properties of the received OFDM/OQAM signal. Moreover, due to the significant computational complexity of the derived ML estimators, a closed-form CFO synchronization algorithm based on the LS method is considered. It is also derived, under the assumption of low SNR, the joint ML symbol timing and phase offset estimator for AWGN channel. Since the phase estimate is in closed form, by substituting its expression in the likelihood function, a blind symbol timing estimator that requires only a one-dimensional maximization procedure is obtained. The ML symbol timing estimator exploitsboth the conjugate and the unconjugate cyclostationarity properties of the OFDM/OQAM signal that are related to the bandwidth of the adopted pulse-shaping filter.The thesis is organized as follows. In Chapter I, an introduction to FBMC systems is provided. The transmitter and receiver for both FMT and OFDMOQAM systems are presented and it is put in evidence the central role of prototype filters very well localized in time and frequency. In Chapter II, the sensitivity of FBMC systems to the presence of synchronization errors is analyzed. Chapter III deals with data-aided synchronization techniques for FBMC systems both in downlink and up-link scenarios. The joint LS CFO and symbol timing estimator for FMT and OFDM/OQAM systems is derived. The joint CFO and symbol timing estimator in multi-user OFDM/OQAM systems is also proposed. In Chapter IV, it is looked at blind synchronization algorithms for FBMC systems. A Blind CFO estimator for noncritically sampled FMT systems is proposed. Moreover, non data-aided CFO estimators for low SNR conditions for OFDM/OQAM systems are presented. The ML symbol timing estimator for low SNR conditions for OFDM/OQAM systems is also derived. Finally, conclusions are drawn in Chapter V.