Some possible suggestions for topics:
One possible way of getting ideas (although some might be
end up being hard: look at recent ISIT conference websites,
(e.g. google ISIT 2006)
and look at the plenary talks and tutorials. These will
give you some ideas of active topics in information theory,
although they often have one of the plenary talks on a
non-information theory subject.
1. Information theory in practice
How do cell phones work? What are FDMA, CDMA, TDMA, coding?
How does CDMA work?
The theory of multiple-access and broadcast
channels, and its relation to cell phones.
How does image compression work?
How does video compression work?
How does the coding in CD's work?
2. Theroetical advances in information theory:
MacWilliams identities and the linear programming bound for
minimum distance in a code.
The work surveyed in
Constrained
sequences, crossword puzzles, and Shannon (you'll need more references
than this, but you should be able to find quite a few of them).
What is space-time coding? How do multiple antennas help information
transmission?
What is the theory (and maybe some history) of turbo codes?
What is the theory (and maybe some history) of LDPC codes?
Explain trellis codes and the Viterbi algorithm. How are trellis codes
used?
What is Slepian-Wolf coding (sec. 15.4)? What are practical codes for
this problem.
Present Lovasz's result about the zero-error capacity of the 5-cycle, and
survey the current state of results on zero-error capacity.
3. Algorithms in information theory:
The Burrows Wheeler transform, and how to do it and its inverse
quickly.
The Berlekamp-Massey algorithm for decoding BCH (and related) codes.
The Arimoto-Blahut algorithm for finding the optimal probability
distribution on the input of a channel (discussed briefly in 10.8 of
Cover and Thomas and much more understandably in
Prof. Gallager's
notes.
List decoding (for Reed-Solomon codes and others; see Madhu Sudan's papers
(his survey paper "List decoding: Algorithms and Applications" and others).