Time

17.06.2021 kl. 12.30 - 16.30

Description

Abstract
Stochastic models of the radio channel are widely used simulation tools due to their simplicity and low computational cost compared to the deterministic approaches. Another appealing feature of stochastic channel models is that they are versatile, i.e., they can be applied to different radio propagation scenarios by simply calibrating their parameters. This is easier said than done as most stochastic channel models have intractable likelihood functions, leaving standard parameter estimation methods inapplicable. The standard approach to this problem involves sophisticated algorithms for extraction and clustering of multipath components, following which, point estimates of the model parameters can be obtained using specialized estimators. However, evaluating performance of the end-to-end calibration method becomes infeasible, not to mention the difficulty in implementing these sophisticated algorithms. In this work, we develop likelihood-free calibration methods, primarily based on approximate Bayesian computation (ABC), to calibrate stochastic radio channel models without multipath extraction and clustering. Our methods utilize the temporal moments, which are frequently used statistics for characterizing the radio channel. We show that the temporal moments are informative about the parameters of stochastic channel models, and hence, can be used to calibrate them. Our main contribution is the development of a general ABC method that can calibrate stochastic channel models with different mathematical structures using the exact same procedure and settings. Exemplified by the Saleh-Valenzuela and the propagation graph model, the proposed method can accurately estimate their parameters in simulations as well as when applied to real channel measurements.

Assessment Committee
Professor, Xuefeng Yin, Tongji University, Shanghai, CHINA
Associate Professor, Klaus Witrisal, Graz University of Technology, Austria
Associate Professor, Beatriz Soret, Aalborg University (Chairman)

Supervisor
Associate Professor, Troels Pedersen, Aalborg University

Moderator
Associate Professor, Carles Navarro Manchon, Aalborg University

Host

Wireless Communication Networks Section, Department of Electronic Systems

Address

ONLINE via TEAMS

Registration Deadline

16.06.2021 kl. 12.00

Register at

cnm@es.aau.dk