Daniel Aronsson

Licentiate Thesis, Signals and Systems, Uppsala University, June 2007

The thesis available in Pdf: 1550KB.

Paper copies of the thesis can be obtained from Ylva Johansson, Signals and Systems Group, Uppsala University, Box 534, SE-75121 Uppsala, Sweden.

Abstract:

Digital communications systems require the receiver to estimate the transmitted bit sequence from a noisy received signal. Estimation is therefore a crucial part in digital communications. Prediction of error rates, on the other hand, is not, but it enables capacity improving techniques in the form of fast link adaptation and opportunistic resource scheduling.

In this thesis, solutions to the estimation and prediction problems are proposed by inferring radio channels that vary rapidly due to the mobility of users. It is crucial not only to produce point estimates and predictions of the channels, but also to take the uncertainty of those estimates into account.

This thesis adopts the Bayesian probability interpretation, which regards probability theory an extension to logic. Orthodox statistics, which considers a probability to be a limiting frequency of an imagined experiment, will in many cases produce only point estimates, whereas the Bayesian method also always produces measures of uncertainty.

Linear state space models are designed for a number of system types, and the Kalman filter is used to infer the time-variant radio channels. The proposed channel predictor is evaluated on a specifig system proposal. It is found that control data aiding the channel estimation and prediction (so called pilot data) should be transmitted simultaneously by all users, and that the distribution pattern of pilot symbols should also be varied over time, in order to achieve a high prediction performance.

Two methods for predicting the bit error rate are proposed. It is shown that although the associated mathematical expressions are somewhat involved, the numerical complexity induced by those is negligible compared to the complexity of the channel predictor.

It is also suggested how the proposed algorithms may be used for evaluation and design of wireless multiuser systems.

Related publications:

Eusipco 2007: OFDMA uplink channel prediction to enable frequency-adaptive multiuser scheduling.

IEEE PIMRC 2007: Kalman predictor design for frequency-adaptive scheduling of FDD OFDMA uplinks.

IST Mobile Summit 2005: Adaptive TDMA/OFDMA for wide-area coverage and vehicular velocities.

IEEE ICASSP 2005: Channel estimation and prediction for adaptive OFDMA/TDMA uplinks based on overlapping pilots.

IEEE VTC-2003-Fall: Channel estimation and prediction for adaptive OFDM downlinks.

Wireless IP project complete publication list.

Prediction of mobile radio channels, Ph.D Thesus by Torbjörn Ekman, 2002.