|
Schemes for Engineers in Research and Development
Research Chairs: Profiles
Professor Bernard Mulgrew FRSE FIET - University of Edinburgh
Selex S&AS/Royal Academy of Engineering Research Chair in Signal Processing
Background
 At
School I enjoyed Mathematics and Physics and
stumbled uncertainly into a degree course at Queen’s
University Belfast in Electrical Engineering by way
of first Mathematics and then Civil Engineering. At
Queen’ s I was fortunate to attend the inspiring
lectures of S.Q.A.M.A. Hussain who introduced me to
the beauty of the “fast Fourier transform” algorithm
– one of the jewels in the crown of Signal
Processing. After graduation I landed on my feet
when I joined the Signal Processing Group at the
then Ferranti Radar Systems in Edinburgh. This group
was lead by John Roulston OBE, whose forceful and
dynamic leadership showed me that this “fast Fourier
transform” was not just an elegant piece of
Mathematics but had many significant real-world
applications. After a 4 year apprenticeship with
John, I returned to the academic life to take a PhD
with the very successful academic team of Colin
Cowan and Peter Grant in the Signal Processing Group
at the University of Edinburgh. There we explored
“adaptive filter algorithms” – signal processing
algorithms that learn for themselves, responding to
get the best out of real-world environments.
Research
What is Digital Signal Processing? Take it one word
at a time. Digital – this means that all the
necessary calculations are done on a computer or
some similar bit of electronics hardware with
associated software. Signal – this describes how any
physical quantity such as voltage or a temperature
changes with time (or space). Processing – now this
is the really interesting bit – once we get
measurements of the Signal into our Digital computer
we want to Process the measurements to extract
useful information e.g. how fast is the voltage
changing or have some aspects of the signal changed
with respect to what was happening earlier to-day?
This Processing is designed and described by Signal
Processing Engineers in terms of mathematical
recipes known as algorithms.
Signal processing is everywhere. You cannot avoid
it. For example, it is at the heart of all mobile
radio systems, facilitating commerce and recreation
across the globe. It is one of the key enabling
technologies for both MP3 players and broadband
access to the internet; allowing us to pump video
images down copper wires that were designed for the
earliest telephone systems. Our old friend the “fast
Fourier transform” continues to confirm its jewel in
the crown status as a major component in digital
television receivers.
What do I do now? My own research as a Professor at
the University of Edinburgh has two strands: (i)
generic algorithm development and assessment; (ii)
application specific signal processing. The former
is about topics such as new ways of separating
signals:
Blanco, D. & Mulgrew, B. (2005), 'ICA in signals
with multiplicative noise', Signal Processing,
IEEE Transactions on, 53(8), 2648--2657.
Or adaptive algorithms that respond more actively to
unknown environments;
Bhatia, V. & Mulgrew, B. (2007), 'Non-parametric
Likelihood Based Channel Estimator for Gaussian
Mixture Noise', Signal Processing., in press
Wei, X.; Cruickshank,
D.G.M.; Mulgrew, B. & Riera-Palou, F. (2007), 'A
Unified Approach to Dynamic Length Algorithms
for Adaptive Linear Equalizers', Signal
Processing, IEEE Transactions, 55(3), 908--920.
The more application specific work tends to be done
in collaboration with companies such as BAE SYSTEMS
and Selex S&AS on both radar signal processing:
Lim, C.; Aboutanios, E. &
Mulgrew, B. (2007), 'Training Strategies for
JDL-STAP in a Bistatic Environment', IET
Proceedings Radar Sonar and Navigation.
and challenging filter design problems:
Loeda, S.; Reekie, H. &
Mulgrew, B. (2006), 'On the design of
high-performance wide-band continuous-time
sigma-delta converters using numerical
optimization', Circuits and Systems I: Regular
Papers, IEEE Transactions, 53(4), 802--810.
Or communications receiver design with Lucent-Bell:
Claussen, H.; Karimi, H. &
Mulgrew, B. (2005), 'Low complexity detection of
high-order modulations in multiple antenna
systems', IEE Proceedings- Communications
152(6), 789--796.
Or in collaboration with colleagues in Medicine:
Unsworth, C.; Spowart, J.;
Lawson, G.; Brown, J.; Mulgrew, B.; Minns, R.R.
& Clark, M. (2006), 'Redundancy of Independent
Component Analysis in Four Common Types of
Childhood Epileptic Seizure', Journal of
Clinical Neurophysiology 23(3), 245-253.
Sometimes fortune favours you and you can find
something that appears generic when addressing an
application specific task. This particular algorithm
has the working title of “in-the-gate detection” and
it can detect sine waves in unknown correlated noise
(Fig 1):
Aboutanios, E. & Mulgrew,
B. (2005),A STAP algorithm for radar target
detection in heterogeneous environments, in
'Statistical Signal Processing, 2005 IEEE/SP
13th Workshop on', pp. 966--971.
 |
 |
Figure 1: In-the-gate, constant false alarm
rate detection (T) for space-time adaptive
processing – no secondary data required. |
The Future
Where is the subject
going? The future is in distributed networks
of sensors interconnected by finite capacity
communications links. Such networks of sensors
(Fig. 2) will provide significant challenges for
the Signal Processing Engineer in extracting,
coding and communicating information across the
network against a backdrop of limited battery life
and limited capacity of the communication links themselves.
This is a major theme of the current collaborative work with Selex S&AS.
|
 |
Figure 2: Acoustic Sensor Network |
|
