Performance Analysis of Pilot-Aided LS and MMSE Channel Estimation Techniques for OFDM Wireless Systems

Abstract

Accurate channel state information is essential for coherent detection and equalisation in Orthogonal Frequency Division Multiplexing (OFDM) systems operating over frequency-selective, time-varying multipath channels. Pilot-aided channel estimation, in which known pilot symbols are periodically inserted among data subcarriers, remains the most widely deployed approach in practical OFDM standards. This paper presents a performance analysis of two classical pilot-aided channel estimation techniques -- the Least Squares (LS) estimator and the Minimum Mean Square Error (MMSE) estimator -- together with a Discrete Fourier Transform (DFT)-based denoising enhancement applied to the LS estimate. A complete receiver block diagram and an estimation-procedure flowchart are presented to describe the signal chain from pilot extraction to data equalisation. The estimators are evaluated through an executed numerical simulation of a 64-subcarrier OFDM system with comb-type pilot arrangement operating over a six-tap Rayleigh fading channel with an exponential power-delay profile, and their Mean Square Error (MSE) performance is compared over a signal-to-noise ratio (SNR) range of 0-24 dB. Results show that the plain LS estimator exhibits the highest MSE due to its sensitivity to noise at pilot positions, the MMSE estimator improves on LS by exploiting second-order channel statistics, and the DFT-based enhancement, which truncates the estimated channel impulse response to the known maximum delay spread, provides the lowest MSE among the three by suppressing out-of-delay-span noise. The paper discusses the practical complexity-performance trade-offs among the estimators and outlines directions for further work, including estimation for doubly-selective channels and sparse-channel compressive-sensing-based estimators.