Interview with John Fonseka

12 June 2013

What are your main research interests?

Our main research interest is in the design of powerful error correcting codes with low decoding complexity. Our research focuses on the design of specialised interleaver permutation functions that result in a significantly increased minimum Hamming distance of the concatenated code. This allows simple component codes to be concatenated to construct much more powerful codes. Such an approach allows simple but powerful codes to be constructed that have relatively low encoding and decoding complexity.   We are also interested in the design and analysis of bandwidth efficient modulation schemes, and combined modulation coding schemes.

What has motivated you to research error correcting codes?

We believe that properly designed block accumulate (BA) codes can be very powerful and can have a very low error floor. Specifically, if the interleaver in a BA code can be properly designed powerful BA codes can be found starting with relatively simple outer block codes. Such codes, due to their much improved error floor variations, can perform better than other known powerful codes including turbo codes and low density parity check codes (LDPCs).

Can you explain the advance that you have reported in your Electronics Letters paper?

In our paper, we propose a novel bit by bit row/column interleaver (BBRCI) design method to systematically design the interleaver in a BA code with the intention of increasing the minimum Hamming distance (MHD). We demonstrate that the proposed BBRCI method can generate BA codes with high MHD starting with relatively simple outer block codes. As a result, the BA codes designed according to the proposed BBRCI method have a much lower error floor.

Which particular applications could benefit most from this work and in what ways?

Due to their improved error floor properties, the BA codes developed according to the BBRCI design rules are suitable for low error rate applications. Such low error rate applications include optical transmissions and data storage.

What are you doing to continue and build on this work?

We are actively searching for good interleaver design methods in serially concatenated codes. We believe that the correlation introduced by the outer code in serial concatenation allows interleaver designs that can generate powerful concatenated codes. By properly designing the interleaver’s permutation function, we can greatly improve the MHD of the concatenation while maintaining the bulk of the interleaver gain. We are also actively investigating ways to reduce the decoding complexity of concatenated codes.

We have already published two other papers related to interleaver design in serial concatenation in the IEEE Communications Letters. In one, we have introduced a novel constrained interleaver (CI) design technique that is capable of achieving the maximum possible product distance while simultaneously achieving interleaver gain close to that with uniform interleaving. In the second paper, with suitable modifications, we have extended the CI technique for serial concatenations with inner recursive codes. Specifically, we have introduced two CI techniques: CI-1 that achieves the product distance dodi; and CI-2 that sacrifices some interleaver gain to increase the MHD up to do2di, where, do and di are the MHD of the outer code and the inner code, respectively.  We continue to investigate interleaver design and applications of CI.

How do you see denoising technology progressing over the next few years?

We would like see the use of powerful error correcting codes with low complexity. Currently, the community is emphasising a lot on LDPCs and turbo codes. Even though they can perform well at moderate error rate applications (around 10-5), they suffer from the error floors and are not quite suitable for low error rate applications. Using properly selected interleaver permutation functions, two or more relatively simple component codes can be selected to meet a given minimum Hamming distance requirement while maintaining the bulk of the interleaver gain, thus providing a robust code with a low error floor.  

Further reading

A PDF version (new window) of this feature article is also available.

Journal content

Cover of Electronics Letters, volume 50, issue 22

View all content

Browse or search all papers in the latest or past issues of Electronics Letters on the IET Digital Library.