This week of GSoC I dealt with a first approach to perform a OFDM parameter estimation. As mentioned in the last blog post, my milestones where altered to develop a OFDM estimation block. The parameters that should be estimated are

- Subcarrier spacing
- FFT length
- Symbol time
- Timing offset
- Frequency offset

The offsets should also be compensated. This week I developed a prototype for a OFDM estimator.

## OFDM Estimation

To perform an estimation of the mentioned parameters, I used an algorithm developed in [1]. This work states to have a good robustness against short cyclic prefix lengths and fading channels in comparison to previous agorithms (for instance [2]).

In this paper, four algorithms are presented. I chose to use the cyclic correlation algorithm, because it is able to perform a blind estimation and seems to have a moderate implementation effort.

I implemented a first python prototype of this algorithm and did some performance tests with different SNRs in AWGN and fading channels.

Currently the algorithm seems to perform nicely. Still, performance is a problem (about 15 secs per estimation). The algorithm is based on a cost function that needs to be maximized. This is currently done with brute force over a set of different parameter values. I’ll proceed with a C++ implementation of that algorithm and then try to implement a more efficient optimizing of the cost function.

## Precalculated Taps

As stated in the last blog post, the Signal Separator block now takes JSON files with precalculated filter taps to save CPU load during runtime. I managed to finish the implementation and calculation logic. Now, a user can choose between calculating taps during runtime or using the ones in the provided file. A small python scripts helps to generate such a taps file.

## ToDo

- Implement OFDM estimator in C++
- Performance test/improvement
- Begin with synchronization algorithm

[1] Bouzegzi, Abdelaziz, Philippe Ciblat, and Pierre Jallon. “New algorithms for blind recognition of OFDM based systems.” *Signal Processing* 90.3 (2010): 900-913.

[2] Yucek, Tevfik, and Huseyin Arslan. “OFDM signal identification and transmission parameter estimation for cognitive radio applications.” *IEEE GLOBECOM 2007-IEEE Global Telecommunications Conference*. IEEE, 2007.