Generalization of Wireless Transceivers
Relevant Publications: ICC '22, ICC '21, TWC '20, VTC '18, 5GWF '18, SenSys '19
Modern wireless systems are increasingly dense and mobile, making the channel highly non-stationary, rendering conventional receivers sub-optimal in practice. Predicting the channel characteristics for non-stationary channels has the distinct advantage of pre-conditioning the waveform at the transmitter to match the expected fading profile. The difficulty lies in extracting an accurate model for the channel,especially if the underlying variables are uncorrelated, unobserved and immeasurable. Our work imple-ments this prescience by assimilating the Channel State Information (CSI), obtained as a feedback from the receiver, over time and space to adjust the modulation vectors such that the channel impairments are significantly diminished at the receiver, improving the Bit Error Rate (BER). V2X communication is used as an example of non-stationary channels to demonstrate the efficacy of this approach. To account for the multivariate, non-stationary channel, a tensor decomposition & completion approach is used to mitigate the effects of transients, sparsity and noise in the CSI measurements.
#1. Non-Stationary Wireless Channels
The propagation environment in modern wireless systems like Vehicle to Everything (V2X), High Speed Train (HST), mobile Massive MIMO and modern mmWave networks are typically inherently non-stationary, due to the distance-dependent path loss, shadowing, delay or Doppler drift and time-varying propagation scenarios. Consequently, such non-stationary channels are particularly difficult to analyze. This nature impacts the reliability and latency of data transmission, which has been validated by various measurement campaigns. Significant prior research exists on channel models for modern non-stationary wireless systems, that typically build on geometric stochastic channel models by capturing the temporal evolution of small-scale fading channel characteristics. The degree of non-stationarity of wireless channels is evaluated using metrics to assess the rate of variation of certain local channel statistics such as variation of the power delay profile, or using a correlation matrix distance measure. However, reliable communication over non-stationary channels is very rare in literature, due to the challenging nature of tracking such channel statistics.
#2. Channel Prediction at Transmitter
Recommending channel characteristics for V2X communication has the distinct advantage of pre-conditioning the waveform at the transmitter to match the expected fading profile. The difficulty lies in extracting an accurate model for the channel, especially if the underlying variables are uncorrelated, unobserved and immeasurable. Our work implements this prescience by assimilating the Channel State Information (CSI), obtained as a feedback from vehicles, over time and space to adjust the modulation vectors such that the channel impairments are significantly diminished at the receiver, improving the Bit Error Rate (BER) by 96% for higher order modulations. To account for the multivariate, non-stationary V2X channel, a tensor decomposition and completion approach is used to mitigate the effects of sparsity and noise in the CSI measurements. Overall, the system is shown to operate with a prediction accuracy of 10−3 MSE even in dense scattering environments over space and time.
#3. Eigenspace Precoding at Transmitter
The time-varying distribution and the existence of joint interference across multiple degrees of freedom (e.g., users, antennas, frequency and symbols) in such channels render conventional precoding suboptimal in practice, and have led to historically poor characterization of their statistics. The core of our work is the derivation of a high-order generalization of Mercer’s Theorem to decompose the non-stationary channel into constituent fading sub-channels (2-D eigenfunctions) that are jointly orthogonal across its degrees of freedom. Consequently, transmitting these eigenfunctions with optimally derived coefficients eventually mitigates any interference across these dimensions and forms the foundation of the proposed joint spatio-temporal precoding. The precoded symbols directly reconstruct the data symbols at the receiver upon demodulation, thereby significantly reducing its computational burden, by alleviating the need for any complementary decoding. These eigenfunctions are paramount to extracting the secondorder channel statistics, and therefore completely characterize the underlying channel. Theory and simulations show that such precoding leads to >10^{-4}× BER improvement (at 20dB) over existing methods for non-stationary channels.
#4. Data-driven Receiver with BER Guarantees
Neural Network (NN) based receivers have seen limited adoption in practical systems due to a lack of explainability and performance guarantees, despite their efficacy as a data-driven tool for physical layer signal processing. In order to bridge this gap in explainability, we present an equivalent NN-based receiver that performs the same optimizations used by classical receivers for symbol detection. Achieving equivalence is crucial to explaining how a NN-based receiver classifies symbols in high-dimensional channels and determining its structure that is robust to the underlying channel with minimum training. We realize this by deriving the risk function that guarantees equivalence, which also provides a measure of the disparity between NN-based and classical receivers. Consequently, this information allows us to derive mathematically tight data-dependent bounds on the bit error rate of NN-based receivers, and empirically determine its structure that achieves minimum error rate. Extensive simulation results show the efficacy of the derived bounds and structure of NN-based receivers for single and multi-antenna systems over a variety of channels.
References
Maqsood Careem
Achieving reliable communication over non-stationary channels is very challenging yet critical, and requires long-term data-driven learning models that are fast, adaptive and evolve over time.