Fully-Digital Blind Compensation of Non-Linear Distortions in Wideband Receivers

Compressive Sensing Technique for Mitigating Nonlinear Memory Effects in Radar Receivers

“…The Volterra series is a powerful black-box behavioural model capable of modelling complex nonlinear memory ef-fects. It is often described as a Taylor series with memory and has a unique linear-in-the-parameters formalisation that makes it ideal for modelling weakly nonlinear RF amplifiers [8]. Mathematically it is described as a sum of multidimensional convolutions and in this paper we employed it in its complex baseband (BB) form [24] for computational efficiency.…”

“…Crucially, the nonlinear models tested in the radar simulation had to fit into the Volterra series framework otherwise the above expression for the NCS gradient (30) becomes invalid. The specific nonlinear models tested in the simulation were therefore: the BBTS model, the BBVS model, the BB Hammerstein model and the BB parallel Hammerstein model [8], [31]. For each nonlinear model listed above, the radar simulator was configured to perform statistical convergence analysis on the NCS algorithm whereby the algorithm's mitigation performance was tested against randomly generated sets of kernel coefficients.…”

“…As long as the active components in the RF receiver remain in the linear regime, while the desired signal and the interference signal pass through the chain simultaneously, then the principle of superposition holds and the task of removing the interference effects can be seen as a linear signal separation problem [6]. In an effort to save costs, modern radar systems are being designed from commercial-off-the-shelf RF components which have significantly poorer linearity specifications than their bespoke counterparts [5], [7], [8]. These modern sensors must therefore operate at the top of their linear regions in order to achieve acceptable performance.…”

“…While little work has been published on the mitigation of nonlinear receiver effects in the radar literature [4], [5], [10], [11], [13]- [15] it is an area of active research in the related field of communications. To save costs it is common for communications systems to operate their receivers in the weakly nonlinear regime and compensate for the nonlinear distortion digitally [5], [7], [8], [16], [17]. By far the most popular of these techniques is nonlinear equalization (NLEQ) where the corrupted signal is passed through an inverse of the forward nonlinearity so that the dominant nonlinear effects are suppressed.…”

“…The memory of a system is intrinsically linked to its bandwidth [8], [20] and with modern radars moving to increasingly wider receiver bandwidths they can no longer be approximated as strictly memoryless systems. It was shown in [11] that as nonlinear memory effects are introduced to the radar receiver, memoryless NLEQ techniques become increasingly decorrelated significantly reducing their effectiveness.…”