Maximum Likelihood Frequency Estimation and Preamble Identification in OFDMA-based WiMAX Systems

Abstract

In multi-cellular WiMAX systems based on orthogonal frequency-division multiple-access (OFDMA), the training preamble is chosen from a set of known sequences so as to univocally identify the transmitting base station. Therefore, in addition to timing and frequency synchronization, preamble index identification is another fundamental task that a mobile terminal must successfully complete before establishing a communication link with the base station. In this work we investigate the joint maximum likelihood (ML) estimation of the carrier frequency offset (CFO) and preamble index in a multicarrier system compliant with the WiMAX specifications, and derive a novel expression of the relevant Cramer-Rao bound (CRB). Since the exact ML solution is prohibitively complex in its general formulation, suboptimal algorithms are developed which can provide a reasonable trade-off between estimation accuracy and processing load. Specifically, we show that the fractional CFO can be recovered by combining the ML estimator with an existing algorithm that attains the CRB in all practical scenarios. The integral CFO and preamble index are subsequently retrieved by a suitable approximation of their joint ML estimator. Compared to existing alternatives, the resulting scheme exhibits improved accuracy and reduced sensitivity to residual timing errors. The price for these advantages is a certain increase of the system complexity

On Maximum-Likelihood Blind Synchronization Over WSSUS Channels for OFDM Systems

“…Then the range of correct estimates is given by (5) As to , it is well known that, due to the cyclic nature of some correlation function in , CP-based estimators can only resolve fractional CFO, i.e., the estimate is limited to a frequency range of one subcarrier spacing, say [20]. The integer part of the CFO results in a shift of the subcarriers, which is usually resolved by correlating with one or more known pilot sequences after the initial blind synchronization [6], [21].…”

“…Then its autocovariance can be written as (6) where E denotes expectation, has been called the delay cross-power density [23] and denotes the Kronecker delta function. If the CP is long enough to cover the maximum delay spread, then is zero except for .…”

“…A data-aided approach in this case may thus entail a substantially longer latency (up to about 5 ms, excluding computation time) and a substantially higher complexity (due to the need to examine several tens or more OFDM symbols) than blind methods. In this and other such cases, therefore, using blind methods can facilitate fast, low-complexity initial frequency and timing synchronization, which can be followed by a data-aided scheme to fine-tune the frequency and timing as well as estimate other parameters (such as the cell identity) [5], [6].…”

Efficient Integer Frequency Offset Estimation Architecture for Enhanced OFDM Synchronization

“…Dopplerinduced CFO [4]) and cognitive radio (switching between multiple frequency bands leads to CFO [15], [16]). Like many recent works [17], [18], [19], this paper is concerned with IFO estimation, either to support large-CFO application scenarios, or to enable cost saving through relaxing strict front-end design criteria. While state-of-the-art IFO estimation methods achieve good performance, they rely on crosscorrelation between received signal and known preamble.…”

Unitary PUMA Algorithm for Estimating the Frequency of a Complex Sinusoid

“…In many applications, such as wireless communications [1], synthetic aperture radar imaging [2]- [3], source localization in radar and sonar systems [4]- [6], and software radio receivers [7], fast and accurate frequency estimation is crucial for proper system operation.…”