Faculty of Engineering, Built Environment and Information Technology
School of Engineering
Department of Electrical, Electronic and Computer Engineering
Selected Highlights from Research Findings
Conventional brake technology in trains relies on air pressure pipes running along the length of the train. However, such technology has significant shortcomings in that it is characterised by time delays, therefore imposing an upper limit on the length of trains.
A technological innovation known as electronically controlled pneumatic brakes (ECPB) represents a substantial improvement over previous systems in that control signals are transmitted via wires instead of pipes filled with air.
This allows fast and accurate electronic control of wagon air brakes, making it possible to extend the length of trains to as much as 10 km. ECPB can be used in conjunction with distributed power (DP), which involves distributing locomotives throughout the train to enhance safety and performance. The remote locomotives are unmanned, and are also controlled electronically.
Spoornet is the first railway in the world to implement ECPB/DP on a significant scale. The ECPB/DP system that is currently in use on its COALlink operations can be described as an “open loop” control system in that the driving system is pre-designed to accommodate track topology.
A major limitation of such a system is that it offers relatively limited control. A train driver can control remote groups of locomotives in only two modes: a leading group and a remote group.
Furthermore, it is virtually impossible for a train driver – who is situated at the leading locomotive – to visualise the position of remote locomotives that can be up to 2.5 km away.
In order to overcome these limitations, a research project was undertaken to develop a closed loop control system. Such a system differs from an open loop system in that it adjusts itself online by taking into account current operating conditions such as the train’s position, composition and speed.
This system, when it is implemented, will make it possible to automatically control all the remote groups of locomotives in an optimal manner.
Prof X Xia
Electrical, Electronic and Computer Engineering
+27 (0) 12 420 2165
xxia@postino.up.ac.za
The challenges of higher spectral efficiency, quality of service, and data rates in wireless systems have stimulated research in multiple-input-multiple-output (MIMO) systems for mobile and fixed wireless communications. Accurate channel modelling – the ability to characterise the wireless environment between the transmitter and the receiver - and space time coding are essential to exploit the opportunity presented by the multi-path structure of the channel.
MIMO wireless systems are characterized by multiple antenna elements at the transmitter and receiver, providing the potetial for increased capacity in a rich scattering environment. A signal is scattered by objects such as buildings, walls and fixed objects and can have many paths to the receiver from the transmitter. MIMO systems operate by exploiting the spatial properties of the mutipath channel, thereby offering a new dimension which can be used to enable enhanced communication performance.
The research project encompasses channel modelling of MIMO channels for indoor and outdoor systems. The researchers have developed and implemented (in hardware and software) a cost effective Wideband 8x8 MIMO Channel Sounder at the University that can excite the channel across an instantaneous bandwidth of 80 MHz and capable of operating across the 2-6 GHz band.
The received data can be mathematically manupilated so as to derive the channel matrix that can be used to characterize and model the system for the specfifed environment. The group has undertaken a measurement campaign at the2.4 GHz and 5.2 GHz carrier frequencies in an indoor environment, using the co-located configuration. From the literature this is the most cost effective 8x8 MIMO wideband channel sounder designed to date.
The system was developed in collaboration with Prof Michael A Jensen and Dr Jon Wallace, Brigham Young University, Provo, Utah, who are leading reseachers in MIMO.
Mr BTJ Maharaj
Electrical, Electronic and Computer Engineering
+27 (0) 12 420 4636
sunil.maharaj@eng.up.ac.za
Transceiver design and miniaturisation is a challenge faced by radio frequency (RF) microelectronic designers globally. It is one of the focus areas of the microelectronics and electronics group of the Carl & Emily Fuchs Institute for Microelectronics (CEFIM).
A transceiver incorporates several sub-systems including oscillators, filters and mixers. Design, development and optimisation of each sub-system have been an international focus of several RF research institutes and in the last two decades, several of the sub-systems have been well researched and designed.
With international research co-operation, it is an intention to integrate our very own South African transceiver chip Integrated Circuit (IC). One research gap that has been identified was the need of miniaturization of an oscillator.
In most transceivers, the oscillators are not placed with the rest of the sub-systems, as they deploy an inductor component with its associated design challenges: size and performance.
Our aims of integrating a fully on-chip transceiver involve incorporation of the oscillator (with a special inductor) - the latter making the approach suited to niche development.
Two kinds of inductor implementations exist: a passive inductor and an active inductor. A passive spiral inductor can be built on a silicon substrate by using the multilevel interconnects that are routinely provided with today’s mainstream silicon fabrication processes.
An active inductor is designed using a set of transistors that are mathematically configured to emulate the behaviour of a passive inductor, for example a set of components that does not include an inductor, yet eventually behaves like one.
While an active inductor consumes more power than its passive counterpart it has the advantage of better performance and a size reduction of a factor of about 1000. This provides the opportunity of substantially improving performance. The next phase included incorporating the active inductor for applications such as frequency synthesis in a transceiver environment.
For the design and simulation efforts of an active inductor in a fractional-/N/ frequency synthesizer, a paper of student Jannes Venter and Sinha was accepted by two international conferences (ISSCC 2005 and DAC 2005) and rated the best paper in the conceptual category of a highly prestigious competition run jointly by these two major US conferences.
The research will proceed to inductor design optimisation and the possibility of developing an active inductor kit for semiconductor foundries and RF microelectronic designers – to ease this essential design component for the future.
Mr Saurabh Sinha
Electrical, Electronic and Computer Engineering
+27 (0) 12 420 2950
saurabh.sinha@up.ac.za
A potentially significant contribution was made in the Human Language Technologies research group. Research findings indicate that information on the durations of phonemes can be used to enhance the accuracy and robustness of systems for speaker identification, and to protect such systems against attacks by recordings.
The work is particularly interesting because it may improve the ability of speaker- identification systems to operate in cross-channel conditions, where training examples are gathered in one condition, for example over a landline telephone, and the system is used in other circumstances, for example over a cellular telephone.
Research on the topic is continuing, with the goal of further improving performance and characterizing behaviour under various circumstances.
Prof E Barnard
Electrical, Electronic and Computer Engineering
+27 (0) 12 420 2981
etienne.barnard@up.ac.za
A new variation was introduced on the capacitive feed probe for patch antennas on thick substrates. It consists of a small circular probe-fed capacitor patch that is situated next to the resonant patch. This configuration can bring about significant savings in terms of manufacturing cost, but also lends itself to a very efficient full-wave analysis.
As such, the main focus of this work was a spectral-domain moment-method formulation, which was specifically developed for the analysis of large, but finite, arrays of these antenna elements. Entire-domain and sub-domain basis functions are combined in an efficient way to minimize the computational requirements, most notably computer memory. It is was shown that, for general antenna array configurations, memory savings of more than 1000 times can be achieved when compared to typical commercial software packages where only sub-domain basis functions are used.
A number of numerical and experimental results were obtained in order to verify the spectral-domain moment- method formulation and to illustrate various applications of the new antenna element.
Prof J Joubert
Electrical, Electronic and Computer Engineering
+27 (0) 12 420 2860
johan.joubert2@up.ac.za
|
