We analyze makespan minimization in a tree data gathering network with limited memory. The network consists of a set of worker nodes, a set of intermediate nodes and a single base station. Each of the worker nodes holds a dataset that has to be transferred to an appropriate intermediate node. The intermediate node processes the dataset and then sends it to the base station. At most one worker can communicate with a given intermediate node at a time, and at most one intermediate node can communicate with the base station at a time. A dataset occupies the intermediate node's memory buffer from the moment when it starts being received until the time when its transfer to the base station completes. The total size of datasets coexisting in the memory of an intermediate node can never exceed its buffer size. The scheduling problem is to minimize the total time of gathering the data. We show that this problem is strongly NP-hard even in the special case when there is only one intermediate node and the communication with the base station takes no time. Heuristic algorithms are proposed and tested by means of computational experiments.
Keywords: scheduling, data gathering networks, limited memory, heuristicsScheduled
TC1 Scheduling 2
June 10, 2021 12:30 PM
1 - GB Dantzig