Abstract
Estimating parameters of stochastic radio channel models based on new measurement data is an arduous task usually involving multiple steps such as multipath extraction and clustering. We propose two different machine learning methods, one based on approximate Bayesian computation (ABC) and the other on deep learning, for fitting stochastic channel models to data directly. The proposed methods make use of easy-to-compute summary statistics of measured data instead of relying on extracted multipath components. Moreover, the need for post-processing of the extracted multipath components is omitted. Taking the polarimetric propagation graph model as an example stochastic model, we present relevant summaries and evaluate the performance of the proposed methods on simulated and measured data. We find that the methods are able to learn the parameters of the model accurately in simulations. Applying the methods on 60 GHz indoor measurement data yields parameter estimates that generate averaged power delay profile from the model that fits the data.
Highlights
S TOCHASTIC models of the radio channel are indispensable tools in the design and analysis of communication and localization systems
We extend our previous work on learning parameters of stochastic channel models using approximate Bayesian computation (ABC) [18] and deep learning (DL) [19], where we had applied the methods on the cluster-based model of Saleh and Valenzuela [3]
We present two machine learning methods based on ABC and DL to calibrate stochastic radio channel models without multipath extraction
Summary
S TOCHASTIC models of the radio channel are indispensable tools in the design and analysis of communication and localization systems. A. APPROXIMATE BAYESIAN COMPUTATION ABC is a likelihood-free inference method that samples from the approximate posterior distribution of the parameters by finding values that lead to simulated datasets from the model that are similar to the observed data. Data and statistics are again simulated from the newly generated population, and the M closest parameter samples are accepted and assigned weights w(t) ∝ p(θ )/φt(θ ), where the division is taken entry-wise This sequence of steps is repeated for T iterations, till the approximate posterior distributions converge. To improve the posterior approximation and speed up the convergence, we combine the aforementioned two methods by applying the regression adjustment step on the accepted parameters after each iteration of PMC-ABC. The sample mean of the accepted samples after Tth iteration, (θ 1(T), . . . , θ M(T)), gives the approximate MMSE estimate
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
