While Shannon already characterized the capacity of point-to-point channels back in 1948, characterizing the capacity of wireless networks has been a challenging problem. The deterministic channel model proposed by Avestimehr, etc. (2007 - 1) has been a promising approach

for approximating the Gaussian channel capacity and has been widely studied recently. Motivated by this model, an improved combinatorial algorithm is considered for finding the unicast capacity for wireless information flow on such deterministic networks in the first part of this

thesis. Our algorithm fully explores the useful combinatorial features intrinsic in the problem. Our improvement applies generally with any size of finite fields associated with the channel model. Comparing with other related algorithms, our improved algorithm has very competitive

performance in complexity.

In the second part of our work, we consider the design and analysis of rate-compatible LDPC codes. Rate-compatible LDPC codes are basically a family of nested codes, operating at different code rates and all of them can be encoded and decoded using a single encoder and

decoder pair. Those properties make rate-compatible LDPC codes a good choice for changing channel conditions, like in wireless communications. The previous work on the design and analysis of LDPC codes are all targeting at a specific code rate and no work is known on

the design and analysis of rate-compatible LDPC codes so that the code performance at all code rates in the family is manageable and predictable. In our work, we proposed algorithms for the design and analysis of rate-compatible LDPC codes with good performance and make the code performance at all code rates manageable and predictable. Our work is based on E2RC codes, while our approaches in the design and analysis can be applied more generally not only to E2RC codes, but to other suitable scenarios, like the design of IRA codes. Most encouragingly, we obtain families of rate-compatible codes whose gaps to capacity are at most 0.3 dB across the range of rates when the maximum variable node degree is twenty, which is very promising compared with other existing results.