We propose a model with low communication cost, where the object with largest value is chose as the representative object, and adjustment factors are used to guarantee that local value of representative object is also the largest one in each remote node.

After SPA was proposed,people have done a lot of works in order to get better accuracy while reducing the communication cost.

In order to reduce the communication overhead during the skyline computation on distributed data stream,an idea based on remote filter is presented and the corresponding theoretical foundation is proved.

The communication overhead in parallel implementation of artificial neural network on a multiprocessor system is analyzed in this paper.

When the system is in high communication overhead,the new scheme overcomes the disadvantage of dynamic load balance scheme and enhances the comprehensive properties of system effectively.

This paper Presents an algorithm to identify parallel tasks in Ada,and proposes the computation of the communication costs matrix.

This technique,through application of synchronization control to sequential loops and changing sequential loops into shared ones, will reduce communication overhead and improve program performance.