A unified approximation scheme for the convolution of independent binomial variables

In this paper, we have developed a unified approximation scheme for computing the convolution of independent binomial variables. The development of the scheme has been based on two approximation methods: Speevak's [4] and ours [3]. The two methods have been compared and their properties studied to understand their advantages and limitations. Our knowledge of these advantages and limitations has been well exploited in developing this scheme. Under the scheme, the two methods can be applied alternatively. A criterion for determining which of them is better for a problem is developed. The scheme is found to be devoid of most of the shortcomings associated with each of the two methods and is also more accurate.