Low-complexity iterative reception of Continuous Phase Modulation (CPM) is addressed using Linear Minimum Mean-Square Error (LMMSE) filtering empowered by Soft Interference Cancellation (SIC). The transmitter emits a convolutionally encoded, true CPM signal with a constant envelope. The receiver exploits Laurent-Mengali-Morelli's (LMM) decomposition to reduce the detection complexity, where the transmitted true CPM signal is approximated as a superposition of a small number of parallel Pulse Amplitude Modulated (PAM) and phase-encoded terms. Pulses and phase-encoded pseudo-symbols arising from the LMM decomposition are treated, respectively, as parallel Inter-Symbol Interference (ISI) channels and Multiple Access Interference (MAI) components. Hence the transmitted signal is regarded to inherently contain controlled ISI and MAI, which are equalized for at the receiver front-end using LMMSE-SIC filters. The receiver back-end comprises a serially concatenated turbo scheme encompassing a Soft Input-Soft Output (SISO) CPM detector and a SISO channel decoder. Simulations conducted on various CPM schemes show that the proposed receiver provides significant complexity reduction for a small performance penalty. Consequently, CPM waveforms with long pulses, which have traditionally been deemed infeasible due to the prohibitive complexity, can now be implemented employing the proposed receiver. Hence substantial spectral side-lobe compaction offered by long CPM pulses can be harnessed.

Large-scale (a.k.a. Massive) Multiple-Input Multiple-Output (MIMO) systems are considered as a strong candidate to meet the exceptionally high spectral efficiency requirements for “beyond 4G” (or commonly termed 5G) wireless communications systems. For such systems, the availability of accurate uplink channel knowledge at the base station is critical to success, particularly in the time-division duplex (TDD) mode, where channel reciprocity is exploited to employ efficient downlink beamforming/precoding schemes. A major obstacle to acquiring such channel knowledge at the base station, however, is posed by the potential uplink pilot interference in multi-cell environments known as pilot contamination. In a recent contribution [1], it is shown that pilot contamination can be sidestepped with the aid of a simple interference management scheduling protocol. Building and expanding on [1], we elaborate on the design of shift-and mutually orthogonal pilots, and propose a multi-band operation to expand the number of users to be serviced in densely-populated areas. More specifically, the proposed design adjusts the transmission bandwidth to exploit the spatio-temporal resolution properties of wideband wireless channels, which, combined with the shift-orthogonality principle, enable to allocate identical frequency resources to a number of closely-spaced users. In addition, the pilot transmission technique presented herein minimizes guard interval overhead in the OFDM context, and can be realized with quasi-constant envelope, maximizing battery efficiency in user handsets.

A method based on the Bivariate Empirical Mode Decomposition (BEMD) is addressed to facilitate radar scene analysis for cognitive radar, building and expanding upon a previous contribution. The method exploits the response of BEMD to the fractional Gaussian character of coherent sea clutter returns. Second-order properties of the intrinsic mode functions are used to form a null hypothesis, which indicates the absence of target(s) if accepted. Extensive experiments on real-world radar data show that the proposed radar scene analysis procedure leads to significantly enhanced statistical separability for target+clutter and clutter-alone data. The results are judged from an information-theoretic perspective using the Kullback-Leibler distance as well as by visual inspection.