Novel statistical criteria for local mean power estimation in wireless coverage prediction

Abstract

Much research on the estimation of local mean power is based on Lee's criteria. However, there are some restrictions for Lee's criteria in practical applications. Aiming at enhancing the wireless coverage prediction efficiently, novel statistical criteria including statistical interval, sampling distance, confidence level and number of measurement runs in wireless field coverage prediction are established. In contrast to previous criteria, in which a non-correlation property between sampling points is assumed, an evaluation equation to determine the lower and upper bound of the statistical interval is proposed. Moreover, no previous work has mentioned the confidence level and the number of measurement runs that will introduce estimation errors in the prediction process. The authors will take these factors into consideration to establish the novel criteria for the setting of the confidence level and the number of measurement runs. These criteria are of importance owing to their contribution to wireless network planning and optimisation. In addition, the novel criteria present reasonable explanations as to why some tighter sampling distances, such as 3, 4 and 6 cm in 900 MHz working frequency, can be adopted in the high speed railway construction in China and in certain European countries. However, such tighter sampling distances do not meet the present requirement for sampling distance in local mean power estimation, which is 26 ~ 37 cm in 900 MHz working frequency. The authors will explain the reason through theoretical analysis and simulation.