Communications Research
This began firmly as industrial R&D naturally stemming from work I did with various companies (such as where I worked or consulted - Ericsson Research, Philips/Simoco Telecom, HMGCC and Tait Radio Communications) morphing eventually into some PhD projects. It is very pragmatic work because my focus is always on practical systems rather than pure theory. Thanks are due to many students, in particular Dr Shiva Prakash Premananda (now a senior R&D engineer at Broadcom) and Dr Erwin Anggadjaja (now a lecturer at Suriya University, Indonesia).
Some of my major contributions relating to communications:
Winning, with my team at Tait Electronics (Group Research), the inaugural IEE Innovation in Engineering Awards in 2005. This global award was highly sought after, with many entries from around the world (mainly UK and USA). Our New Zealand based team beat many huge and famous companies to win first prize in the Product Design category. What had we done? We had built the worlds most spectrally efficient narrowband radio system.
Later we would patent this and a spin-off company was formed to commercialise it: MiMOMax Ltd
Inventing, investigating and characterising a cross-layer method to distribute packets along parallel links to take advantage of imbalanced BER characteristics between links.
Note: I have to admit that communications and wireless are not really my greatest skills – I never really studied this stuff at university so it's all self-taught.
Breakdown of my current research (with paper references – note that the best papers are highlighted in blue)
Transmit
|
Transmit Antenna Selection (TAS) is very simply explained: the receiver 'listens' to the received signal and tells the transmitter which antenna (out of a set of possible antennas) that it should use for the next transmission frame. Unfortunately there is a feedback delay – it can take some time for the receiver to 'tell' this to the transmitter and for the transmitter to act upon it. This means the system is often acting upon outdated information... When the channel between Tx and Rx has a low time correlation then it changes rapidly and this means that information is totally outdated. Therefore we need to design channel predictors. The above diagram is a later variant of the system which adds AM (Adaptive Modulation) to the system: the modulation depth, and hence channel data rates, are also adapted in line with channel variations. The work is quite mathematical... but the results include some very pragmatic solutions that are useful to real systems. Mostly it is by ex-PhD student Shiva Prakash Premananda.
Note: I am no longer pursuing this research at the current time, but there is a lot more to be done here by the right research team. |
|
Relaying
|
Again this is very pragmatic work – we consider transmitting from a source to a destination using non-regenerative relays (i.e. amplify-and-forward schemes). Mainly we use MRC at any multi-antenna receiver and with source-selected relays. In this work, my ex-PhD student Shiva Prakash Premananda considers the use of relays. In particular situations with single and multiple antennas at source, destination and relay, with multiple relays are all considered. In fact our papers cover every possible permutation of these systems! Note: I am no longer pursuing this research at the current time, but many other (and better) researchers remain active in this field. |
|
Communications |
This monster is called the STAR platform (see here). It stems from work I did at Tait Radio Communications “Group Research” (and was a collaborative effort of our entire team). This wasn't the first system like this that I built... but it was the biggest! Basically I set my mind to looking at “computation for wireless” in general, including (but not limited to) FPGA-based systems for wireless. Quite a lot came out of this research, and it's still going today thanks to my collaboration with Prof. Suhaib Fahmy (Nanyang Technological University) and PhD student Pham Hung Thinh. |
|
Imbalanced
|
The figure above plots data that I recorded from a real dual-channel VHF digital radio link, at the receiver. It plots the actual BER of each channel instance (i.e. BER over the duration of a packet). The top plot is for 0dB and the bottom is for 5dB SNR average channel noise level. The two channels are plotted back-to-back (i.e. above and below the centre line, which represents no-error), and mean BER for each channel is given as a dotted horizontal line. This is real data, not simulated. The scale is unimportant. Note two things here: 1) It is quite rare for any single packet to actually achieve anything like mean BER. Most packets are either zero error, or three times mean BER... 2) It is even less likely that both channels will be very good simultaneously, or that both will be very bad simultaneously. In fact, this type of data distribution is very common in real wireless systems, but very uncommon in simulated wireless. Most academic papers and simulations just calculate “what is the BER achieved at a mean SNR”. Very few consider the fact that most of the time a wireless system does not experience mean SNR! So, over a number of years, I set my mind to thinking how we can build a pragmatic system that is agile enough to respond to changing channel conditions. One output of this was the TAS research mentioned above (although TAS went far, far, beyond this). Another output was cross-layer systems performing packet switching, which I even extended into the field of WSNs with various students (mainly Dr Anggadjaja).
|
|
© 2014 by Professor Ian McLoughlin of NELSLIP and USTC.