This paper explores the recovery and rate capacity effect for batteries used in embedded systems. It describes the prominent battery models with their advantages and drawbacks. It then throws new light on the battery recov ery behavior, which can help determine optimum discharge profiles and hence result in significant improvement in battery lifetime. Finally it proposes a fast and accurate stochastic model which draws the positives from the earlier models and minimizes the drawbacks. The parameters for this model are determined by a pretest, which takes into account the newfound background into recovery and rate capacity hence resulting in higher accuracy. Simulations conducted suggest close correspondence with experimental results and a maximum error of 2.65%.
A major constraint in design of mobile embedded systems today is the battery lifetime for a given size and weight of the battery. With the tremendous increase in the computing power of hardware and the relatively slow growth in the energy densities of the battery technologies, estimating the lifetime and energy delivered by the battery has become increasingly important to choose between alternative implementations and architectures for mobile computing platforms. Currently, designers of mobile computing systems, while using traditional energy optimization approaches like DVS  tend to assume that the battery is ideal, that is, it would have a constant voltage throughout the discharge, and would also have a constant capacity for all discharge profiles, which is not always true.