Maximum Likelihood Time Synchronization for Zero-padded OFDM

TL;DR

This paper develops an optimal maximum-likelihood time synchronization method for zero-padded OFDM, which is more power-efficient than cyclic prefix OFDM, and demonstrates its superior performance through simulations.

Contribution

It derives a closed-form joint PDF for ZP-OFDM samples and proposes an efficient NDA-ML synchronization algorithm using Monte Carlo sampling and golden-section search.

Findings

Outperforms existing algorithms in lock-in probability

Effective in frequency-selective fading channels

Suitable for low-power wireless systems

Abstract

Existing Orthogonal Frequency-Division Multiplexing (OFDM) variants based on cyclic prefix (CP) allow for efficient time synchronization, but suffer from lower power efficiency compared to zero-padded (ZP)-OFDM. Because of its power efficiency, ZP-OFDM is considered as an appealing solution for the emerging low-power wireless systems. However, in the absence of CP, time synchronization in ZP-OFDM is a very challenging task. In this paper, the non-data-aided (NDA) maximum-likelihood (ML) time synchronization for ZP-OFDM is analytically derived. We show that the optimal NDA-ML synchronization algorithm offers a high lock-in probability and can be efficiently implemented using Monte Carlo sampling (MCS) technique in combination with golden-section search. To obtain the optimal NDA-ML time synchronization algorithm, we first derive a closed-form expression for the joint probability density function (PDF) of the received ZP-OFDM samples in frequency-selective fading channels. The derived expression is valid for doubly-selective fading channels with mobile users as well. The performance of the proposed synchronization algorithm is evaluated under various practical settings through simulation experiments. It is shown that the proposed optimal NDA-ML synchronization algorithm and its MCS implementation substantially outperforms existing algorithms in terms of lock-in probability.