Otimização na seleção de alvos em uma interceptação telefônica

Abstract

In this study, a way to model the relations between members of a criminal organization who have judicial authorization to break telephone secrecy and registration data was proposed. The motivation for this proposal came during Operation Flashback I, in which several duplicate calls were noticed caused by the strong interaction between the intercepted targets, which leads to a waste of human (time) and computational resources (such as the storage of calls on the server, the allocation of the channel that diverts the telephone call to the branch, etc.). For the purpose of minimize such effects, the relation between monitored individuals were represented using graphs from reverse extracts provided by telephone operators. In the modeling, the vertices of the graphs represent the interlocutors and the edges represent the calls made. Each edge contains a weight that represents the total hours of calls from individuals who communicated. From the relation graph, a variation of the problem of coverage of vertices to the graph was used to obtain the suggestion of a set of individuals to be monitored in the following interception cycle. This variant, known as k-maximum vertex coverage, takes as input a weighted graph and an integer k, the objective of the problem is to find the k-coverage (subset with k vertices) that maximizes the sum of the edge weights. Therefore, considering that this problem is NP-complete, an algorithm that solves this problem in polynomial time is not known, however, there are strategies that do it in viable time. In this work, a solution was designed based on the mathematical formulation of integer linear programming. With the proposed solution, it was possible to find optimal solutions in real and synthetic instances (up to 50 times larger than the real ones) with a low computational cost. Accordingly, succinctly, the contributions of this work include (i) a mathematical model for the selection of targets to be monitored, (ii) a way to numerically quantify how much of the network formed by the criminal organization and its interlocutors is being monitored and (iii) ) an algorithm that maximizes the weight of the k-coverage of vertices of the graph formed by the monitored individuals and their interlocutors, whose objective is to suggest individuals to be monitored in the following interception cycle.