Dragan Urošević

Variable neighborhood search (VNS) is a framework for building heuristics based upon systematic changes of neighborhoods both in a local search phase, to find a local minimum, and in a perturbation phase to escape from the large valley. In this chapter, we provide an overview of different VNS variants and describe how they can be used to solve some variants of the Traveling salesman problem. More precisely, we present different neighborhood structures that may be exploited, and show how they can be organized within variable neighborhood descent and variable neighborhood search heuristics. Finally, we provide insights on the performance of different VNS methodologies applied to two variants of the\linebreak Traveling~salesman~problem.

OR-Library is a platform that provides standardized examples for testing problem-solving algorithms in many fields of the operational research and combinatorial optimization. One of the problems emerged in the previous decade is the p-next center problem. The solution to the p-next center problem implies locating p centers in order to minimize the maximum user distance to the closest center plus the distance between that center and the center closest to it. There are several heuristic algorithms for solving this NP-hard problem that give optimal or near-optimal solutions. We propose an algorithm for solving the p-next center problem based on the variable neighborhood search method, capable of recognizing whether some of the found solutions from the OR-Library test set are exact. As a result of the algorithm execution, more than 50% of the solutions are identified as globally optimal. The paper presents a table with the found exact solution values for the p-next center problem from the OR-Library test set.

This paper considers the problem of data access control when the subscribers are IoT devices with initialization that cannot be updated during the entire life cycle. A generic framework and a particular instance for conditional data access control within IoT are proposed. The generic framework is based on the employment of a dedicated secret key-based broadcast encryption scheme where encrypted credentials for conditional data access is available in the blockchain and encrypted data subject to conditional access are available in an off-chain source of streaming data. Reduction of the keys management overhead in comparison with a straightforward decryption keys delivery is experimentally illustrated. An instance of the proposed framework built over the Ethereum blockchain platform is developed and experimentally evaluated.

The p-center problem is a well-known and highly studied problem pertaining to the identification of p of the potential n center locations in such a way as to minimize the maximum distance between the users and the closest center. As opposed to the p-center, the p-second center problem minimizes the maximum sum of the distances from the users to the closest and the second closest centers. In this paper, we propose a new Variable Neighborhood Search based algorithm for solving the p-second center problem. Its performance is assessed on the benchmark instances from the literature. Moreover, to further evaluate the algorithm?s performance, we generated larger instances with 1000, 1500, 2000, and 2500 nodes and instances defined over graphs up to 1000 nodes with different densities. The obtained results clearly demonstrate the effectiveness and efficiency of the proposed algorithm.