Unique-Word-Based OFDM

In the foreseeable future OFDM (orthogonal frequency division multiplexing) will be one of the most important wireless communication technologies. In OFDM subsequent symbols are separated by guard intervals to prevent intersymbol interference. In these guard intervals usually cyclic prefixes (CPs) are inserted. In this project a different OFDM transmission concept, where the CPs are replaced by deterministic sequences, that we call unique words (UWs), shall be investigated. Since the UWs are known deterministic sequences, they can advantageously be used for system parameter estimation purposes at a receiver. In recently accepted conference papers and one journal submission we showed how UWs can in principle be generated in OFDM symbols. The proposed approach can be interpreted as a complex number Reed-Solomon (RS)-code structure within the sequence of the OFDM subcarriers. Redundancy is introduced by appropriately loading a set of so called redundant subcarrier symbols. At the receiver we suggested an LMMSE (linear minimum mean square error) estimator that acts as a combined zero forcing and noise reduction stage. The overall system outperforms classical CP based OFDM (CP-OFDM) in typically frequency selective environments, even though suboptimum data estimation approaches have been applied in our preliminary work. The following research directions shall be focused in depth within the research project: Design and generation of the UW: In our current approach we reduced the total energy of the redundant subcarrier symbols on average by the introduction of a permutation matrix. Nevertheless, the redundant energy is still in the same magnitude as that of the data symbols. Different ways to further reduce the redundant energy shall be studied. First investigations also showed that our transmit scheme features an improved spectral sidelobe behavior compared to CP-OFDM. This aspect shall be investigated analytically, and additional possible optimizations, e.g. concerning the permutation matrix, shall be analyzed. Finally, the UW itself has to be designed properly. On the one hand spectral requirements shall be fulfilled, and on the other hand the UW shall be designed to support system parameter estimation procedures in an optimum way. UW-based system parameter estimation: In wireless digital communication systems a number of system parameters have typically to be estimated at the receiver side. The research objectives are to investigate the applicability of the UW for carrier frequency offset estimation and tracking, I/Q amplitude- and phase-offset estimation, and channel frequency response estimation and tracking. Data estimation using UW-based a priori information: The LMMSE estimator is a suboptimum but rather low complex solution for a UW-OFDM receiver. On the one hand complexity and performance optimizations of the proposed linear concept shall be studied, on the other hand non-linear estimation concepts like sphere decoding or non-linear MMSE concepts shall be investigated. Peak to average power ratio (PAPR) reduction methods: Like any OFDM signalling scheme, UW-OFDM suffers from a high peak to average power ratio. PAPR reduction methods proposed for classical CP-OFDM shall be adapted and optimized for UW-OFDM.

Orthogonal Frequency Division Multiplexing (OFDM) is currently the dominating digital transmission technique in most of the modern broadband wireless and wired communication systems. One of the drawbacks of classical OFDM is that up to 20% of the transmit energy is wasted for so called guard intervals, which are placed between successive OFDM symbols to cope with the effects of multipath propagation. These guard intervals are built by random, data dependent sequences. In Unique Word OFDM (UW-OFDM), which has been invented by the involved project partners, the guard interval is built of a deterministic - the so-called unique word. In contrast to previous attempts with deterministic sequences in the guard interval, the addressed UW-OFDM signaling approach introduces a certain level of redundancy in the frequency domain, which can be exploited by the receiver in order to improve the system performance. In this project a number of different methods for the generation of UW-OFDM symbols have been introduced and investigated. The most effective approach turned out to be the so called non-systematically encoded UW-OFDM. Its main advantage is the fact that it is able to reach a specified performance with significantly reduced mean transmit power compared to conventional signaling concepts like classical OFDM, an aspect which is crucial for battery-driven devices. Furthermore, non-systematically encoded UW-OFDM shows significantly lower out-of-band radiation compared to classical OFDM, which is a major benefit as the occupied frequency spectrum is an expensive resource. Besides the generation of UW-OFDM symbols we also investigated a number of linear as well as nonlinear data estimation concepts. For most of the successive investigations the best linear data estimator has been used, since it turned out, that nonlinear data estimators allow only minor improvements in the bit error behavior at the cost of significantly increased complexity. Finally, the robustness of UW-OFDM against channel estimation errors and carrier frequency offsets has been studied. The work and results of this project have also motivated other research groups to work on several aspects of UW-OFDM, and because of the presented and convincing properties this transmission technique is now seen as a potential waveform for future broadband communication systems by a number of researchers as well as industry representatives.