Aerospace Science and Technology 29 (2013) 287–295 Contents lists available at SciVerse ScienceDirect Aerospace Science and Technology www.elsevier.com/locate/aescte Using an effective tabu search in interactive resources scheduling problem for LEO satellites missions Arezoo Sarkheyli a,∗ , Alireza Bagheri b , Bahman Ghorbani-Vaghei c , Reza Askari-Moghadam d a Computer Science Department, Payam Noor University of Tehran, Iran b Computer Engineering and IT Department, Amirkabir University of Technology, Tehran, Iran c Railway Engineering Department, Iran University of Science & Technology, Tehran, Iran d New Science and Technologies Department, University of Tehran, Tehran, Iran article info abstract Article history: Received 22 November 2011 Received in revised form 9 November 2012 Accepted 15 April 2013 Available online 22 April 2013 Keywords: LEO satellite Tabu search heuristic Resources scheduling problem Graph coloring theory Resources scheduling in Low Earth Orbit (LEO) satellites is an important optimization problem because of the satellites’ specific constraints. This article addresses a scheduling problem for LEO satellites missions to assign resources which could be satellites or ground stations to the most number of requested tasks by considering the tasks’ priority and satisfying temporal and resource constraints. In this study, first, the scheduling problem is modeled using the graph coloring theory. Then, a new tabu search (TS) algorithm is applied to solve the problem. The proposed algorithm employs a new move function to enhance the exploration ability. Accordingly, an attempt is made to compare the result of the proposed TS with some well-known optimization algorithms. The computational results denote the efficiency of the proposed algorithm, as well as its ability to find schedules that are guaranteed to be near-optimal.  2013 Elsevier Masson SAS. All rights reserved. 1. Introduction The main mission in a LEO satellite is taking images of specific areas of the ground based on the users’ requests. In addition, there are some requested tasks for making connection between satellites and ground stations to renew the satellite’s on-board and power capacity. Since these kinds of requests may be satisfied by several satellites or ground stations, the problem is not separable by satel- lites. In addition, satellite-specific constraints, satellite priorities of certain payload and special operations, as well as visibility conflicts have to be considered in scheduling. These constraints make satel- lite scheduling more complicated than usual Job-Shop scheduling. Consider a set of satellites and a collection of tasks which must be processed by the satellites within a predefined time period. Two types of satellites’ tasks are considered in this study: mission and supporting. As the main purpose of LEO satellites is taking images of specific areas, image acquisition from an area of the earth is named mission task, such as “Area A1 must be imaged for 2 minutes between 09:00 and 09:15”. In addition, supporting tasks are employed to complete satellites missions. For example, down-linking is a task to replenish capacity of renewable resources (e.g. on-board memory), such as * Corresponding author. E-mail address: arezo.sarkheyli@gmail.com (A. Sarkheyli). “Satellite S1 must downlink to a ground station for 5 minutes between 11:07 and 11:31”. In general, supporting tasks are more complex than mission tasks. As each satellite has a limited memory to store the images, they need to communicate with ground stations periodically in or- der to download satellite images [23]. Since a satellite does not need to communicate with one specific ground station, the tasks have to establish a connection between that satellite and a feasi- ble ground station. Consequently, each supporting task is defined for one specific satellite and any ground station which has appro- priate conditions for accomplishing the task. The theory of scheduling is the study of allocating a list of resources over time to perform a collection of tasks [17]. In the recent years, different types of scheduling problems for LEO satel- lites have been considered by a variety of researches. For example, Globus et al. [10,11] presented initial results of a comparison of evolutionary and other optimization algorithms such as simulated annealing and genetic algorithms. They found that simulated an- nealing outperforms the others. Also, Frank et al. [8] proposed a constraint-based algorithm which employed a stochastic search heuristic. In addition, Wolfe and Sorensen [30], Parish [21], Man- sour and Dessouky [20] employed a genetic algorithm to create near-optimal schedules for LEO satellites. But Wolfe and Sorensen [30] showed that their algorithm performs much slower than other well-known algorithms. Agnese et al. [1] showed this scheduling problem is a difficult combinational optimization problem which could be viewed as an instance of a valued constraint satisfaction 1270-9638/$ – see front matter  2013 Elsevier Masson SAS. All rights reserved. http://dx.doi.org/10.1016/j.ast.2013.04.001