DISTRIBUTED FUNCTIONS ALLOCATION FOR RELIABILITY AND DELAY OPTIMIZATION.

The authors present two optimization algorithms for allocating the functions of a given distributed task so that the reliability is maximized and the communication delay is minimized. Distributed task reliability measures the probability of executing successfully a task that is composed of a set of functions running on remote processing elements, while task delay describes the average delay incurred during the processing of a task. The allocation of resources such as computing functions and file influences both reliability and delay. Two different approaches are developed to allocate the resources of a given task. In the first method, the problem is reduced to a 0-1 integer linear programming problem and can be solved as two subproblems. In the second approach, a compound objective function is constructed to measure both reliability and delay.