On linear MMSE based turbo-equalization of nonlinear volterra channels

Abstract

OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible.

On linear frequency domain turbo-equalization of non linear Volterra channels

“…As discussed in [10], the time varying exact MMSE implementation is computationally prohibitive. This encourages investigating some low complexity implementations, among which we select the No-Apriori implementation.…”

“…This encourages investigating some low complexity implementations, among which we select the No-Apriori implementation. Recalling the results in [10], the MMSE solution writes with the NA approximation as:…”

“…This approximation was referenced as implementation-b in [10] and was previosuly proposed in [16]. The equalizer outputx n ∼ N (Cx n , v en ) and the output extrinsic LLR L e The a posteriori LLRs are made extrinsic (L e ) by subtracting the a priori probability fed by the decoder: L e (c n,i |x n ) = L ap (c n,i |x n ) − L a (c n,i ).…”

“…In this work, we use the frequency domain Volterra channel model, to derive an iterative linear MMSE FDE. As argued for the time domain solution in [10], time domain linear equalization is able to cope with both linear and non linear interference when used in an iterative way. Similarly, the proposed iterative FDE is able to cancel both linear and non linear soft symbols which asymptotically approaches the Additive White Gaussian Noise (AWGN) ISI-free performance.…”

“…Indeed, new equalization trends are based on iterative receivers for their excellent Shannon-bound approaching abilities [7], [8], and [9]. In [10], we presented an iterative time domain MMSE-based equalizer to mitigate the channel linear and non linear interference. Different low complexity approximations were investigated.…”