Modeling MIMO channels using a class of complex recurrent neural network architectures

Sarma, Kandarpa Kumar; Mitra, Abhijit

doi:10.1016/j.aeue.2011.08.008

Cited by 27 publications

(21 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These channel coefficients are generated using the standard Clarke-Gans model with parameters given in Table 2 and as described in [15,16] and are combined with OFDM symbols generated using the parameters shown in Table 1. While the signal propagates through the channels, significant and CCI coefficients get involved.…”

Section: Resultsmentioning

confidence: 99%

“…The training and estimation time required, however, is much less than the pure ANN based systems as described in [15,16]. It can also deal with timevarying channels but there is always a scope for further improvement.…”

Section: Fuzzy Time Delay Fully Recurrent Neural Network (Ftdfrnn) Bamentioning

confidence: 96%

“…The results show significant improvement in processing time and accuracy as compared to ANN based approaches. The fuzzy systems, on an average, provide at least 5% improvement in accuracy as compared to the RNN-based estimation which has already been established to be a better alternative to statistical and ANN based approaches [16]. This is in addition to the processing time advantage that the fuzzy systems provide.…”

mentioning

confidence: 89%

“…The core of a FRNN is a fuzzy recurrent neuron (FRN) shown in Figure 4 (b). The MLP block of a FMLP is replaced by a TDFRNN [16] with split-activation so that the combination can use de-coupled in-phase and quadrature components of input signals and use a SOM to combine the outputs in an optimized form. The output of such a set of neurons for a set of inputs X j and state vectors u N;j is expressed as…”

Section: Fuzzy Time Delay Fully Recurrent Neural Network (Ftdfrnn) Bamentioning

confidence: 99%

See 3 more Smart Citations

Recurrent Fuzzy-Neural MIMO Channel Modeling

Sarma

Mitra

2012

Journal of Intelligent Systems

Self Cite

View full text Add to dashboard Cite

Fuzzy systems and artificial neural networks (ANN), as important components of soft-computation, can be applied together to model uncertainty. A composite block of the fuzzy system and the ANN shares a mutually beneficial association resulting in enhanced performance with smaller networks. It makes them suitable for application with time-varying multi-input multi-output (MIMO) channel modeling enabling such a system to track minute variations in propagation conditions. Here we propose a fuzzy neural system (FNS) using a fuzzy time delay fully recurrent neural network (FTDFRNN) that has the capability to tackle time-varying inputs in fuzzified form and is used to model MIMO channels. The inference engine is constituted by novel FTDFRNN blocks which determine the decision boundaries and tracks in-phase and quadrature components of input signals encompassing stochastic behavior of the MIMO channel. The system shows significant improvement in performance compared to statistical and ANN approaches in terms of faster processing time, lower bit error rate (BER) margins and better precision while carrying out symbol recovery of transmitted data through severely faded MIMO channels.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Fuzzy Time Delay Fully Recurrent Neural Network (Ftdfrnn) Bamentioning

confidence: 96%

mentioning

confidence: 89%

Section: Fuzzy Time Delay Fully Recurrent Neural Network (Ftdfrnn) Bamentioning

confidence: 99%

See 2 more Smart Citations

Recurrent Fuzzy-Neural MIMO Channel Modeling

Sarma

Mitra

2012

Journal of Intelligent Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…Artificial Neural Networks (ANN) have been used for system identification [1] [2][3] because these can learn fro m the surrounding environment and model free data and use knowledge for subsequent processing. Different ANN architectures have received attention for modelling and analysing a range of co mmun ication channels [4][5] [6]. A NN finds its widespread application due to its ability to approximate any nonlinear mapping between inputs and outputs given a hidden layer with nonlinear activation function.…”

Section: Introductionmentioning

confidence: 99%

Nonlinear model based prediction of time varying SISO-MIMO channels using FANN-DFE combination

Bhuyan

Sarma

2013

2013 1st International Conference on Emerging Trends and Applications in Computer Science

Self Cite

View full text Add to dashboard Cite

In this paper, we present a method for real-time identification and prediction of time varying mobile radio channel. The predictor comprises of two main units. First, a Feed Forward Artificial Neural Network (FANN) is used to identify the auto regressive (AR) system underlying the fading mechanism. Estimated AR model is then used to predict the future state of the channel. The so obtained channel state information (CS I) is used to equalize the received symbol when it arrives. Since training symbols used are minimal in the prediction state, a Decision Feedback Equalization (DFE) type of architecture is proposed to aid the correction step. The model identification is done for channel gains generated by Jack's model. Experimental results show that the coupled predictor performs better compared to the model or the DFE alone in time varying S IS O-MIMO channels.

show abstract