Supervised Learning of Channel Delay Spreads for Cyclic-Prefix Free OFDM Systems

Author(s):  
Jami Hema Ganesh ◽  
Jen-Ming Wu
2019 ◽  
Vol 0 (0) ◽  
Author(s):  
Asmaa Benieddi ◽  
Sid Ahmed Elahmar

AbstractDirect detection optical orthogonal frequency division multiplexing (DDO-OFDM) systems for a long-reach of standard single mode fiber (SSMF) require a large length of cyclic prefix (CP) to avoid the inter-symbol interference (ISI) effect caused by group velocity dispersion (GVD). Unfortunately, this method is inefficient due to the energy wasted in CP samples. In order to reduce the CP length and to mitigate the residual ISI, a novel blind adaptive channel shortening equalizer (CSE) is proposed in this paper. Based on the orthogonality between subcarriers in the fast Fourier transform (FFT) property, the proposed algorithm attempts to minimize the sum-squared correlation (SSCM) between each sample located in a well-defined window to update the CSE coefficients. Thus, the combined channel-CSE response is shortened. Therefore, it can cancel the residual ISI effect due to the GVD and the short CP length. The performance of the system is evaluated on basis of bit error rate (BER) versus optical signal to noise ratio (OSNR) for different CP lengths. The simulation results validate the new algorithm SSCM and show that it can reduce the CP length with a much better system improvement than existing algorithms.


Sign in / Sign up

Export Citation Format

Share Document