For Intelligence and Higher Spectrum Efficiency: A Variable Packing Ratio Transmission System Based on Faster-than-Nyquist and Deep Learning

1 Aug 2020  ·  Peiyang Song, Fengkui Gong ·

With the rapid development of various services in wireless communications, spectrum resource has become increasingly valuable. Faster-than-Nyquist (FTN) signaling, proposed in the 1970s, has been a promising paradigm for improving spectrum utilization. This paper tries to apply FTN to parameter-variable communications and presents an intelligent high-spectrum-efficiency transmission system based on FTN and deep learning (DL). In the proposed variable packing ratio (VPR) transmission, the transmitter adjusts the packing ratio of the FTN stream at every specific time according to the channel condition information (CSI) or other strategies to guarantee an acceptable bit error rate (BER) or other performance targets. And benefit from the proposed DL-based estimation, the receiver can individually estimate the packing ratio of current FTN transmission without any dedicated channel or extra time slot to carry its information. Besides, the transmitter can also employ a random packing ratio at every pre-specific time to improve the security of the transmission. To demonstrate the capability of the proposed scheme to achieve a higher spectrum efficiency (SE), in this paper, we derive SEs of the proposed VPR transmission under different channels, which are also available for conventional FTN signaling. The numerical results show the correctness of the derivation and the SE gains of the VPR scheme to conventional Nyquist transmission. Also, in this paper, we present a simplified DL-based symbol packing ratio blind estimation and employ it in our proposed VPR system. Results show that the proposed simplified estimation achieves nearly the same performance as the original structure with a dramatically reduced complexity.

PDF Abstract
No code implementations yet. Submit your code now



  Add Datasets introduced or used in this paper

Results from the Paper

  Submit results from this paper to get state-of-the-art GitHub badges and help the community compare results to other papers.


No methods listed for this paper. Add relevant methods here