Staggered Time and Frequency Search to Aid Frame Synchronization

Abstract

This paper tackles the problem of frame synchro- nization in the presence of timing and frequency uncertainty. Considering a discretization of the timing uncertainty domain in time slots, and of the frequency uncertainty domain in frequency bins, this novel technique, called staggered time and frequency search, exploits a joint controlled single dwell Threshold Crossing criterion in both domains. In fact, differently from the conven- tional techniques, this method does not require any parallelism, but digitally controlling the sampler and the oscillator is in charge to scan different hypotheses in time and frequency until the acquisition is completed. According to the analytical and simulated results shown in this paper, this novel technique can obtain performance that is close to the one that can be achieved with the classical techniques, with complexity and power consumption reduction. Synchronization is a key issue in TDM/TDMA systems, since it is the necessary prerequisite for correct data demodu- lation and decoding. Conventionally, frame synchronization is based on a data-aided approach that has to detect the start of a known preamble, identified as Unique Word (UW), which marks the start of the entire frame. This operation is typically divided in two phases in cascade: acquisition and tracking. Acquisition is in charge of getting a first rough estimate, while tracking is asked to possibly detect erroneous synchronization events. This paper addresses the problem of frame acquisition, which is typically the most critical phase because it has to be performed in the presence of unknown carrier parameters and large uncertainty regions, up to the entire frame duration at the terminal cold start. In order to reduce complexity, the uncertainty region is usually discretized in one or more hypotheses per symbol, so transforming the epoch estimation problem in to a detection problem, where the correct hypoth- esis H1 has to be selected among many wrong hypotheses H0. In this scenario, the two main impairments that have to be considered for a robust design are frequency errors, due to oscillator mismatches and Doppler effects, and non-ideal sampling, which introduces useful energy degradation and inter-symbol interference when discretizing the uncertainty region. These two impairments are responsible for a strong degradation of the detection performance making acquisition a critical issue in the system design. Considering a discretization of the timing uncertainty do- main in time slots, and of the frequency uncertainty domain in frequency bins, acquisition can be seen as a two-dimensional problem that translates into the exploration of a bidimensional matrix which can be scanned serially or in parallel. The