On the optimal frame size of linear prediction techniques

Loss of data in Communication Systems is a major problem. The main sources of data loss include congestion, sensor faults, mismatch of measuring devices and limited memory of the buffer registers. Three linear prediction (LP) techniques i.e. Normal Equation method, Levinson Durbin Algorithm (LDA) and Leroux Gueguen Algorithm (LGA) are used to predict the missing data. The objective of this paper is to analyze the three linear prediction techniques for both internal and external prediction and the effect of frame size on different parameters i.e. computational time, prediction gain and prediction error is also being analyzed. The LP algorithms are modified and a threshold limit for prediction error is set to keep the error bounded. A sub-optimal frame size is selected for all the three LP methods and in the end the three algorithms are compared for efficient performance.