Modulation classification in wireless OFDM systems with MAP estimator: Masterarbeit

Adaptive modulation is a technique witch increases the bandwidth efficiency of orthogonal frequency division OFDM systems communication on time-varying multipath prolongations channels. The communciation systems applies bandwidth efficient modulation schemes on subchannels with a high channel quality and robust modulations on channels with bad channel quality. The potential increase of performance due to adaptive modulation is maximal if the adaptive modulation is subcarrier based. The receiver needs knowledge of the adaptive modulation, for the purpose of demodulation. The common technique is an explicit signaling of the adaptive modulation. Especially the discussed subcarrier based adaptive modulation requires a lot signaling overhead. An approach to avoid the signaling overhead is automatic modulation detection (AMC), where the receiver is able to blindly recognize the adapted modulation format without additional signaling. In this Thesis three AMC algorithms are proposed. Two of the algorithms are solution for a classification problem without any information about the transmitter. Under such circumstances, maximum likelihood modulation detection provides the optimal solution in the sense of minimizing the detection error probability. However, the maximum likelihood detection generally requires very high computational efforts. To enable practically feasible solutions, the computational complexity has to be reduced but in such way that the hardware implementation is enabled and at the same time the detection reliability can be maintained. The proposed algorithms only have a moderate increased implementation complexity compared to the common approximations, but it outperforms them significantly. The increase of performance is reached by incorporation more constellation points than the common approaches. The Q-Detector, the main achievement of this thesis, is a maximum a posterori (MAP) modulation detector. The algorithm is designed to classify adaptive modulation allocated by the algorithm of Fischer and Huber. The Q-Detector combines the information extracted out of the signal itself and prior knowledge of the adaptive modulation scheme. Due to the super position of the power allocation, included in the algorithm of Fischer and Huber, and the physical channel, the estimated channel at the receiver is quantized. The Q-Detector exploits the quantized structure in several ways to extracted the prior information. For a further decrease of the classification error probability the transmitter performs an additional adaptive phase shift. Due to adaptive power allocation and adaptive phase shift, the Q detector has a nearly optimal classification performance in the sense of minimizing the additional package error caused by the AMC algorithm. Despite the superposition of power allocation and physical channel, a bidirectional adaptive modulation scheme is proposed at the end of the thesis