A physical-layer network coding (PNC) scheme based on MIMO linear detection is proposed in this paper for a two-way relay channel with a multi-antenna relay. We apply a real-valued linear combiner to the output symbols of the MIMO linear detector to obtain a superimposed constellation for which the labelling is obtained from the exclusive or operation on the two transmitted symbols. The superimposed constellation is QAM-like which can facilitate the SER (symbol error rate) analysis.

Interleave-division multiple-access (IDMA) is a multiple-access technique, which uses an elementary signal estimator (ESE) to achieve simple chip-by-chip detection. In this paper, we propose a new ESE which can be calculated directly based on the two-dimensional received signals. Hence, the proposed ESE can be applied to IDMA using signal modulation with either Gray labelling or non-Gray labelling. In contrast, the original ESE is calculated either based on the one-dimensional received signals or separately based on the real parts and the imaginary parts of the two-dimensional received signals. In fact, the original ESE can only be applied to IDMA using BPSK or Gray-labelled QPSK. From the proposed ESE, we can predict that the IDMA using the non-Gray labelling will be superior to the IDMA using the Gray labelling at the high signal-to-noise ratio (SNR) region. With EXIT (extrinsic information transfer) chart analysis, we can estimate the SNR region for which there is advantage of using non-Gray labelling for IDMA. The simulation results fit well with the theoretical analysis.

We propose a multiple-candidate shaping technique for turbo coded modulation systems. Shaping gains are obtained by encoding each message into multiple turbo coded modulation sequences, called candidates, and then selecting the candidate with the lowest energy for transmission. Multiple candidates are generated by adding various scrambling sequences to each message sequence. The proposed shaping technique can also be applied to turbo coded modulation systems equipped with some other shaping methods to obtain additional shaping gains.

Predicting an individual customer's likelihood of purchasing a specific item forms the basis of many marketing activities, such as personalized shopping recommendation. Collaborative filtering and association rule mining can be applied to this problem, but in retail supermarkets, the problem becomes particularly challenging because of the sparsity and skewness of transaction data. We present HyPAM (hybrid Poisson aspect model), a new probabilistic graphical model that combines a Poisson mixture with a latent aspect class model to model customers' shopping behavior. We empirically compare HyPAM with two well-known recommenders, GroupLens (a correlation-based method), and IBM SmartPad (association rules and cosine similarity). Experimental results show that HyPAM outperforms the other recommenders by a large margin for two real-world retail supermarkets, ranking most of actual purchases in the top ten percent of the most likely purchased items. We also present a new visualization method, rank plot, to evaluate the quality of recommendations.