User talk:N4HY

From DttspWiki

N4HY Talk

09 July 2008:

I apologize for this being my first entry but it will not be my last. Frank has a vision for how things are to proceed for DttSP 3.0. That is good and I like the ideas. My role in this will continue to be what it has been. That is implementation of pieces of the machinery that make it do specific tasks and to optimize the performance in as many senses as possible. But certainly the quality of the output signal of the machinery and the computational complexity are number 1 and 2. That is my comfort zone in this work. I like to argue with Frank over details of the vision but he is definitely the right person to be the system architect.

My first bit of work, which I hope to undertake will enable a piece of Frank's vision and simultaneously provide a revolutionary step forward in most people's thinking about SDR. Alex Shovkoplyas, VE3NEA and fred j harris (San Diego State University and an adjunct member of our staff at work) have provided both the inspiration and technical know how to take what I have wanted to do and make it a reality. CWSkimmer shows what can be done, with low computational complexity, when you apply the well known and widely understood Multirate signal processing apparatus. fred harris has published a terrific book on the subject. In addition, he has made the techniques for designing the details in such a system in an optimize way for the task at hand. I will build a subsection here to publish this and other technical work that will be undertaken in the near future.

04 August 2008:

I spent five days with Frank in Vancouver and we laid out some thoughts. I came back to Princeton and worked on and off for the next 3 weeks on this and it compiles. It makes big time use of the gnu scientific library and its vector and matrix functions. It is not yet complete (nor functions) but it will soon. The simplicity of the code itself is what is so remarkable about the very powerful and clever ideas in multirate systems. I spent much of the last ten days working with fred harris and I have been studying fred's book. Having him here for the summer and available long term as an adjunct is going to be valuable. fred and I have worked out a wavelet like decomposition of a wideband sampled system where you have a lower bound on the channel size. We have figured out the optimal structure for doing this with the ability to use overlapping bands when the signal of interest is not contained in one of the bands. We have figured out how to (in step one) pay a penalty of doing twice as much halfband work as the simplest dissection of the band to gain the ability to do perfect reconstruction disassembly of the band AND eliminate aliasing from the smaller subcomponents. We can also peel off larger signal bands during the process. We believe for those systems where you have a well defined smallest band, this will be a great boon to building complex arbitrary bandwidth receivers using the polyphase filter bank approach with wavelet like deconstruction of the band using halfband filters. I am extremely excited about this work. More to follow.