Fast Algorithms For Fbmc And Gfdm In Dynamic Spectrum Access

Dynamic spectrum access or cognitive radio is widely recognized as a major technology in the coming fifth generation (5G) communications to greatly increase spectral efficiency. In cognitive radio, different networks/systems may dynamically share some spectrum to achieve the maximum overall capacity. Each network/system may just use a few subcarriers or a few non-contiguous frequencies at a given time. It is very challenging to design the waveform to effectively use the fragmented spectrum with high flexibility for dynamical change. Filter bank multicarrier (FBMC) and generalized frequency division multiplexing (GFDM) are regarded as good candidates for the waveform at such situations. In this paper, we propose fast algorithms to implement the FBMC and GFDM when a number of cognitive radio users sharing the same spectrum. The major contributions are as follows. (1) We use the "sparse" property of the user's spectrum to reduce the sampling rate and complexity in the discrete time implementation; (2) We propose fast algorithms for computing the special Fourier transform, which reduce the complexity by nearly half.