Applications of Multiple Element Antennas to Enhance Performance of Wireless Communication Systems

School of Information Technology and Electrical Engineering The University of Queensland, Brisbane, Qld, Australia, 4072 Email: meb[at]itee.uq.edu.au

Abstract: One of the main problems faced by future wireless communications is the rapid increase in the demand for broadband services and applications. Because the available frequency spectrum is limited, high data rate communications in multi-user scenarios has to be achieved using extra means in addition to conventional frequency/time and code multiplexing. The only extra degree of freedom provided to the wireless communications system designer is the space or angular domain. Utilization of this domain requires the use of multiple element antennas (MEA) with appropriate signal processing algorithms. This presentation concentrates on two specific areas of applications of MEA to enhance performance of wireless communications systems, which are researched at the University of Queensland. The first area concerns Multiple Input Multiple Output (MIMO) systems, in which MEA are used both at the transmitter (TX) and receiver (RX) to increase the transmission capacity or to enhance diversity. The other area concerns Multiple Input Single Output/Single Input Multiple Output (MISO/SIMO also known as smart antenna) systems operating over a wide frequency band. With respect to the first area, we concentrate our efforts on developing suitable signal propagation models which can be useful for assessing the MIMO system capacity and its performance when various space-time coding schemes are applied. To this purpose, we work on two types of models. The first model is a simple but fully accurate Electromagnetic (EM) model, which takes into account EM interactions within array antennas as well as scatterers. The model which is currently under development includes parallel wire dipoles which form both TX and RX antennas as well as scatterers. This MIMO system can be studied under static or random conditions. In the undertaken approach the operation of MIMO system is compared to a wave guiding structure, which features multi-mode operation. Using this analogue, advantages and limitations of MIMO can be deduced in advance. The second model takes into account full EM interactions within TX and RX antennas. However, interactions with scattering objects are assumed using a single bounce ray approach. The scatterers being uniformly distributed in an assumed area (representing an indoor or outdoor environment) do not affect the operation of TX or RX array antennas. In our considerations we include both Non-Line of Sight (NLOS) and LOS signal components whose composition is characterized by power ratio (Rician factor). Using this model the effect of LOS component on operation of MIMO system is investigated. By comparing results obtained with the full EM and semi-EM model, the limitations of the latter can be identified. During presentation, results for the MIMO system capacity as a function of inter-element spacing, arrays orientation and the presence or lack of LOS component will be given. With respect to wideband MISO/SIMO systems, we concentrate on the application of a redundant set of antennas to eliminate delay networks and filters that are required in conventional wideband beamformers. We show that wideband beam forming in azimuth can be achieved using a rectangular array of wideband antenna elements and an over sampled Inverse Digital Fourier Transform applied to signals at the array elements. The resulting signal weighting coefficients are real-valued and in practice can be realized using amplifiers or attenuators. During the presentation, we will demonstrate the operation of this wideband spatial beam forming method in examples of small and medium size arrays.

Fading Channels: What can we achieve without "perfect side
information" ?

Australian National University and NICTA

Abstract: The analysis of  fading channels  is often performed under the
assumption that the channel state information (CSI) is perfectly
known to the receiver and/or transmitter. These cases are well
understood by the research community. However, for some wireless
channels, CSI may be difficult to estimate. In such scenarios
where neither the transmitter nor the receiver have CSI, the
channel is not well understood for MIMO or even SISO systems. This
talk will give a brief introduction to this problem and highlight
some of the novel results obtained at NICTA/ANU.

Performance of Space-Time Coded OFDM Systems With Frequency Offset in Fading Channels

ARC Special Research Centre for Ultra-Broadband Information Networks

An Affiliated Program of National ICT Australia (NICTA) Department of Electrical and Electronic Engineering University of Melbourne, Vic 3010, Australia

Abstract: tba

Truncated Maximal Ratio Combining for Iterative Multiuser Decoding

UNISA

Abstract: In this paper, the concept of information combining is investigated for iterative multiuser decoding. This is done by applying recursive maximal ratio combining (MRC) over iterations to interference cancellation based multiuser detectors. Here, we show that combining over all iterations may not be appropriate as correlations build up. Numerical results show that applying MRC onlyover the first few iterations can improve the bit error rate performance with only a modest increase in complexity, as compared to no combining.

Detection guided adaptive estimation

School of Information Technology and Electrical Engineering The University of Queensland, Brisbane, Qld, 4072 Email: homerj[at]itee.uq.edu.au  

Abstract: In many adaptive estimation applications the system being estimated has a sparse parametric representation. For example, in
the application of acoustic echo cancellation, the impulse response of the echo path often consists of only a 'small' number of active(non-negligible) response regions interspersed by inactive (negligible) response regions. Other sparsely parametrised examples include: multi-path characteristics within mobile phone and digital television transmissions; the spatial distribution of user signalswithin a cellular radio cell; the cross-talk between individual DSL lines in a VDSL cable.

Despite the sparse nature of such systems, many commercial adaptiveestimation algorithms adaptively estimate the active as well as the
inactive parameters. This approach is popular because it requires minimal a-priori knowledge of the system. On the other, as implied by Vapnik and Chervonenkis learning theory and other dimension dependent studies, significant gains should be achieved by adaptively estimating only the active parameters. Motivated by this, we have recently developed a least squares (training-based) active parameter detection algorithm, which may be incorporated in many (training-based) adaptive estimation algorithms, such as the NLMS.

Simulation studies have demonstrated that such least squares based detection guided adaptive estimators show significantly improved convergence and tracking capabilities and/or steady state performance. Importantly, the computational complexity of the detection guided adaptive estimators is only marginally greater than that of the standard adaptive estimators.

In this seminar we describe the theoretical basis of the least squares based detection algorithm and propose/investigate extensions of this algorithm to a range of (training-based and blind) adaptive applications, including: equalisation of mobile communication channels; smart antenna array processing for cellular CDMA systems; cross-talk suppression within VDSL cables.

Temporal Change in MIMO Systems

Department of Electrical and Computer Engineering University of Canterbury

Abstract: Models for the time varying nature of a SISO wireless channel are well-known and are largely based on the Jakes process. In MIMO systems, however, it is often the case that information about the individual channel coefficients are not required. Instead, various functions of the MIMO channel, such as eigenvalues, eigenvectors, condition numbers, etc. are more useful. Hence, the question is whether standard models for the time variation of  individual channel coefficients can be used to obtain results for a variety of channel functions. We discuss an approach to solving such problems based on stochastic differential equations which govern the temporal evolution of Brownian matrices. 

This is work in progress with Min Zhang, Mansoor Shafi, Lee Garth, Iain Collings and Matthew MacKay.

Performance Analysis of Adaptive QAM Schemes with Non-zero Round-trip Delay

WATRI

Abstract: Link adaptation techniques have recently been proposed as a spectrally efficient method of obtaining high quality service for mobile communication systems. These schemes aim to better utilise channel capacity compared to fixed transmission schemes, by adapting signal transmission parameters such as modulation constellation and transmit power. Adaptive modulation schemes, which adapt the modulation constellation, have gained considerable favour for exploiting time-varying channel conditions without increasing the level of co-channel interference. However, the traditional adaptive modulation schemes are designed assuming zero delay between the channel estimation and the modulation mode adaptation. In the case of non-zero delay, the transmitter would update the modulation mode based on outdated channel state information from the receiver. In this presentation, an adaptive quadrature amplitude modulation
(AQAM) scheme is proposed and described. By way of retransmissions, this scheme allows a targeted reliability level to be met, even in the presence of non-zero delay. The effect of non-zero delay on performance is then investigated. For this purpose, expressions are derived for the average number of bits per symbol (BPS) throughput and the average bit error rate (BER) of the proposed scheme, and of a conventional AQAM scheme that does not use retransmissions. Numerical results obtained using the aforementioned expressions are then presented. The proposed retransmission based scheme is demonstrated to achieve a significantly lower average BER than the conventional scheme in the non-zero delay case.

Turbo list-detection for multi-user channels

Agere Systems Australia

Abstract: The turbo principle is able to approach the performance of the optimal receiver in many multi-access systems with manageable complexity. Many approaches have been proposed for the CDMA multi-user detector part of the algorithm, which represents the computational bottleneck of the receiver. Most perform poorly when the number of users significantly exceeds the number of observations.

In this work we present approaches to the problem based on the so called "iterative list detection" technique. The methods are shown via simulation results to support very high loads with excellent performance and low complexity.

Throughput Performance of Low Complexity MIMO Extensions to OFDM-Based WLANs

The University of Sydney

Abstract: This talk considers two low complexity MIMO extensions for OFDM-based wireless local area networks (WLANs). The first is a spatial multiplexing (SM) approach employing a linear zero forcing (ZF) receiver; the second is based on space-time block codes (STBC). The throughput performance is demonstrated within the context of IEEE 802.11a. Low complexity channel estimation is employed using orthogonal STBC training matrices. We show that at low SNRs the STBC approach yields the best performance, whereas for higher SNRs, and for channels exhibiting significant delay spread, the SM-ZF scheme offers the better solution. Furthermore, the SM-ZF scheme is shown to perform close to SM-ML. Based on the results, a low complexity hybrid scheme is proposed which switches between SM-ZF and STBC to achieve high throughputs over the entire range of SNRs.

Zarko B. Krusevac 

Differential Coding for Time Varying Binary Symmetric Channel

NICTA and UNSW zarko.k[at]student.unsw.edu.au

Abstract: The classical information theory widely applied in wire-line communication systems over (nearly) time invariant communication channels needs to be updated significantly in order to give answers to a growing number of questions related to the theory and practice of modern wireless mobile communications systems over time varying multidimensional channels. Some of these questions are as basic as " what is the maximum information rate that can be transmitted over an unknown time variable fading channel with arbitrarily small probability of error? "

In 1948 Shannon demonstrated that by proper encoding of the information, errors of the information induced by a noisy time invariant channel can be reduced to any desired level without sacrificing the rate of the information transfer. The classical (Shannon) information theory implicitly assumes that the communication channel parameters are known or at least externally measurable with a negligible capacity loss (overheads) incurred to the overall communication capacity. If the channel is (nearly) constant then this assumption is very much justified. If the channel is time varying, then there is an additional (statistical) process at work, namely the channel (state) process.
Consequently, frequent channel parameter readjustment by additional estimation is necessary. Since all channel estimation algorithms use channel resources for training, explicitly or implicitly (blind), the channel process violates the basic implicit assumption under which the Shannon coding theorem is derived.

The channel process entropy (or the entropy of the channel as an information source) is the amount of information produced by the channel process. This entropy exists independently of the actual information transmission. Since the observation of the channel process is affected by the channel noise, the channel entropy in presence of observer (or the amount of information that one needs for the channel estimation) is not necessarily the same as the channel process entropy. Thus, the question is: Does the channel process need to be estimated before the signal detection takes place or could the signal detection and channel parameter estimation be conducted jointly by proper encoding?

In principle, communication over a time varying channel is possible at any rate below the channel capacity [Gallager]. However, good maximum likelihood (ML) coding strategies for time varying channels (channels with memory) are difficult to determine and the decoder
complexity grows exponentially with memory length [Gallager]. Thus, a common strategy for channels with memory is to disperse the memory using an interleaver and then apply coding techniques for memoryless channels. Nevertheless, this approach achieves a lower inherent Shannon capacity than the original channel, since coding is restricted to memoryless channel codes. The complexity of ML decoding can be reduced significantly without this capacity degradation by implementing a decision-feedback decoder [Goldsmith]. However, the impact of error propagation is ignored. That means, the perfect channel state estimation (based on a "noisy" observation) is assumed.

We use an original approach to investigate information theoretic aspects of time varying communication channels and to provide insight into issues of optimal information transmission. We present the time varying binary symmetric channel (TV-BSC), which is a time varying equivalent for the binary symmetric channel. Then, we use the TV-BSC to confirm that the separate approach to the channel parameters estimation and data detection (almost all existing applications) is not optimal. Finally, we show that the cascade of the differential encoder, the TV-BSC channel and differential decoder can be used for the information transfer over the TV-BSC without capacity degradation. This approach can be extended to the more general class of the Finite State Markov Models for MPSK signaling, which has very strong practical foundation. Thus, although the analysis is basically related to the TV-BSC, still significant conceptual inside in optimal information transfer over more general time varying communication channels can be gained.

(Joint work with Predrag B. Rapajic UNSW Sydney, NSW 2052 p.rapajic[at]unsw.edu.au and Rodney A. Kennedy NICTA and ANU Canberra ACT 0200, rodney.kennedy[at]anu.edu.au )

High Speed Downlink Packet Access reception: hardware implementation challenges and directions.

Agere Systems Australia

Abstract: This talk will introduce the High Speed Downlink Packet Access (HSDPA) mode of the Universal Mobile Telephony System (UMTS) Third Generation (3G) standard. This is designed to deliver high speed data services (up to 14Mbps). Numerous new technologies are introduced to the air interface to achieve higher rates. This provides a significant challenge when designing a chipset to receive HSDPA channels in a user device. In this talk I will outline some of the challenges, solutions and touch upon some of the advanced techniques under consideration for further optimisation. A recurring theme is the significant constraints imposed on a design through the practicalities of hardware implementation. I will talk about the role of the Agere Systems Australia research and development team within Agere's global effort.

Optimal Spectrum Balancing for Digital Subscriber Lines

School of Information Technology and Electrical Engineering The Univ. of Queensland

Abstract: Crosstalk is a major problem in modern DSL systems such as ADSL and VDSL. Static spectrum management, the traditional way of ensuring spectral compatibility, employs spectral masks which can be overly conservative and lead to poor performance.

In this talk we present a centralized algorithm for optimal spectrum management in DSL. The algorithm uses a dual decomposition to solve the spectrum management problem in an efficient and computationally tractable way. The algorithm shows significant performance gains, typically doubling the achievable data-rate when compared with existing spectrum management techniques.