A container bin packing algorithm for multiple objects

Abstract

The bin packing problem considered in this paper has the character with items arriving in batches, such as the task dispatching process in the cloud. The problem has two packing objects, one of which is to minimize the number of used bins, and the other is to reduce the packing time. To meet the objects, a container packing algorithm is presented, which consists of rounds of packing. In one round the number of the necessary bins is first calculated, and those bins are opened in batch to pack items in parallel. This round will be iterated until there are no unpacked items. The proposed algorithm is compared with other classical bin packing algorithms through experiments, which have the same property of simplicity for implementation. It is shown that the novel algorithm not only saves the packing time, but also achieves better combined performance than others.