A Comparison Study of Kalman Filter and H∞ Dynamic Observer for Adaptive Estimation of Time-Varying Radio Channel Quality

Visiting researcher: Dr Xuewu Dai (UCL)
Host researcher: Prof. Yang Yang (WiCO)
Duration: 2 months
Status: Completed
 

Biography of the visiting researcher

Dr. Xuewu Dai is currently a research associate with the Communications and Information Systems research group at UCL. Prior to that, he conducted his PhD study at the internationally reputational Control Systems Centre in the University of Manchester (Oct. 2005 – Dec. 2008) examining “Observer-based Parameter Estimation and Fault Detection”. In this study, he investigated the robust observer (filter) for modeling and condition monitoring of complex dynamic systems (e.g. gas turbine engines). He has published about 20 refereed journal/conference papers and his research interests include the parameter estimation/optimisation, observer-based estimation and filtering, robust condition monitoring, wireless networked control system, medium access control (MAC) delay analysis and communication system simulation.
 
 
Xuewu received his Msc degree in Computer Science and BEng in Communication Engineering both from Southwest University (China) and was awarded the UK/China Scholarships for Excellence 2005. He served as reviewer for several IEEE Transactions and top conferences.
 
 

Background and Motivation

Using Single carrier frequency division access (SC-FDMA), which will be use in Long-Term Evolution (LTE) systems, in time-varying multipath fading channels has been challenging the deployment of the 3GPP systems for a very high velocity application, such as high speed trains. The time-varying multipath fading destroys the subcarrier orthogonality and introduces the inter-carrier interference (ICI). In order to mitigate the ICI in OFDM-like systems, there have been some reports on the application of minimum mean square error (MMSE) estimation to channel quality estimation. Most of them assumed a time-invariant channel and only a few dealt with the time-variant channel.
 
From a system’s viewpoint, the estimation of channel quality is a problem of estimating unmeasureable variables from noise-corrupted observations, which has been studied intensively in the field of control and system identification. Inspired by the recent development of robust estimation techniques (e.g., H∞ observers) for condition monitoring of dynamic systems, it is of interest to apply these new techniques to see if a better solution can be found to the problem of channel estimation subject to time-variant frequency-selective multipath in high velocity applications. As a pilot study that has never been proposed, it is essential to evaluate the performance and make a comparison to these existing methods. The main objectives are to get the test-bed data at WiCO’s channel emulator, develop a novel channel quality estimator by applying the Kalman Filter and H∞ Dyanmic observer and to evaluate and compare their estimation performance on the test-bed data.
 

Research Activities

  1. Build a model of the SC-FDMA signal transmission over the radio channel of 2.4-2.6GHz subject to time-varying channel. Base on WiCO’s EB Prompsim C8 Radio Channel Emulator, four kinds of radio channel model have been built.
  2. Build a SC-FDMA test system and measure the received signal. A QAM transmitter (Agilent E4433B RF signal generator), a radio channel (EB Prompsim C8) and receiver (an Agilent Vector Network Analysisor) are connected to simulate a SC-FDMA transmitter-receiver system. The transmitter send the training sequence (a Zadoff-Chu (ZC) sequence) periodically at a fix interval.
  3. Build a MMSE (Minimum Mean Square Error) and a Kalman Filter to estimated the channel parameters from the measurements of received signal.
  4. Write a technical report on test system, experiment results and the feasibility of H∞ dynamic observer for adaptive estimation of time-varying channel quality.
  5. Throughput the visit, one telephone conference every week between the host institute WiCO and the University College London to report the progress and discuss technical issues.

Outcomes

  1. A technical report on the experiment and feasibility study.
  2. A joint paper is planned to be submitted to a journal.
Powered by Drupal, an open source content management system