Identifying sufficient deception in military logistics

• Published in: Expert systems with applications Vol. 141; p. 112974
• Main Authors: Ausseil, Rosemonde, Gedik, Ridvan, Bednar, Amy, Cowan, Mark
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 01.03.2020; Elsevier BV
• Subjects: Anti-access/area-denial; Commodities; Constraint modelling; Deception; Decision making; Greedy heuristics; Integer programming; Linear programming; Logistics; Military; Military logistics; Mixed-integer linear programming; Optimization; Production scheduling; Shipments; Transportation; Warfare; Anti-access/area-denial; Greedy heuristics; Deception; Mixed-integer linear programming; Military logistics
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2019.112974

•An expert system to incorporate deception in military logistics is proposed.•A non-convex mathematical model is used to mimic adversary and logistics decisions.•Two greedy heuristics are proposed to solve realistic problem instances.•Greedy heuristics are more efficient than the commercial solver.•Sufficient deception levels for different transportation scenarios are quantified. In the context of modern warfare, military forces are faced with challenges that threaten their power projection operations. Adversarial strategies could hinder the actions of the forces within operational areas by inferring and anticipating a military’s potential mission objectives including military logistics and transportation activities. Thus, this research investigates the potential ways of incorporating two military deception strategies into the military logistics decision making process: (i) falsifying signifies including empty convoys on routes on which commodities are being hauled, and (ii) hiding refers to concealing routes in the logistics network. Assuming that the adversaries can monitor the shipments in the transportation network, we develop a two-stage integer linear programming model that quantifies the sufficient amounts of each deception needed to deceive the adversaries. In addition, the mathematical model accurately captures several other military logistics challenges, such as satisfying the demands for military commodities at destinations subject to limited capacities at supply points, deciding the best route option, and respecting the time limit constraint. However, the two-stage model is non-convex and therefore, reformulation and linearization steps are applied to convert it to a convex single-stage model. Due to the computational complexity in solving the single-stage model for realistic-size instances, two greedy heuristics are developed that can efficiently solve those instances. Moreover, the computational experiments reveal that the greedy heuristics obtain near optimal solutions when compared to the solutions of the smaller instances solved to optimality via a commercial solver.