Luigi De Giovanni

Optimization models for the single delay management problem in public transportation

Passengers travelling in public transportation networks often have to use different lines to cover the trip from their origin to the desired destination. As a consequence, the reliability of connections between vehicles is a key issue for the attractiveness of the intermodal transportation network and it is strongly affected by some unpredictable events like breakdowns or vehicle delays. In such cases, a decision is required to determine if the connected vehicles should wait for the delayed ones or keep their schedule. The delay management problem (DMP) consists in defining the wait/depart policy which minimizes the total delay on the network. In this work, we present two equivalent mixed integer linear programming models for the DMP with a single initial delay, able to reduce the number of variables with respect to the formulations proposed by the literature. The two models are solved by a branch and cut procedure and by a constraint generation approach respectively, and preliminary computational results are presented.

Efficiency and Robustness in a Support Platform for Intelligent Airport Ground Handling

One of the effects of the development of air traffic during recent years has been the increase of congestion on major airports and their aprons, where ground handling operations take place. Efficiency and safety issues become important on the aprons. The “Integrated Airport Apron Safety Fleet Management—AAS” project aims at setting a decision support system for efficient and safe management of apron traffic, taking advantage of information gathered by “intelligent” vehicles equipped with on-board positioning and monitoring systems. In this project, we optimize the assignment of vehicles to apron operations. We present a mathematical formulation of the problem and sketch a fast sequential heuristic, which provides efficient assignments based on real-time data on position and status of vehicles and operations. We then discuss how the procedure can reach good trade-offs between efficiency and robustness against unpredicted, but frequent, delays occurring in real time at airports. The proposed approach has been applied to a major European airport (Berlins Tegel), leading to efficiency improvements and reduction of expected delays due to handling operations.

Optimal Client-Server Configuration of Mobile Ad-Hoc Networks

Abstract Interactive online gaming is a widely successful application that is becoming more and more popular even on mobile devices. In this context, an emerging scenario is represented by having playersʼ mobile devices directly connected to each other, without resorting to the Internet. To enable this scenario, we leverage on a hybrid architecture that allows client-server games to be played in ad-hoc mode: a number of players take turn in acting also as the server of the game. However, the problem on how to choose the servers so as to maximize the duration of the network still remains an issue. To this aim, we propose an Integer Programming model and a heuristic based on it: preliminary experimental results show the effectiveness and the potential of the approach.

Optimal shift partitioning of pharmacies

The pharmacy service requires that some pharmacies are always available and shifts have to be organized: a shift corresponds to a subset of pharmacies that must be open 24hours a day on a particular week. Under the requirement that each pharmacy belongs to exactly one shift and the assumption that users minimize the distance to the closest open pharmacy during each shift, we want to determine a partition of the pharmacies into a given number of shifts, such that the total distance covered by users is minimized. It may be also required that shift cardinalities are balanced. We discuss different versions and the related computational complexity, showing that the problem is NP-hard in general. A set packing formulation is presented and solved by branch-and-price, together with a fast solution technique based on a tabu search. They have been applied to real and random instances showing that (i) the set packing formulation is very tight and often exhibits no integrality gap; (ii) the branch-and-price solves problems of practical relevance to optimality in a reasonable amount of time (order of minutes); (iii) the tabu search finds optimal or near-optimal solutions in order of seconds.

Optimal shift coloring of trees

Abstract This paper was motivated by the problem of scheduling the openings of pharmacies during week-ends and holiday periods (shifts). The problem can be modeled as a coloring problem on a graph. In this paper we focus on the special case where the underlying graph is a tree, or, more generally, it is endowed with a tree-metric, and we provide a polynomial-time algorithm. We also provide direct optimal solutions for special trees like stars and paths.

Optimal configuration of active and backup servers for augmented reality cooperative games: Optimal configuration of active and backup servers for augmented reality cooperative games

Interactive applications as online games and mobile devices have grown in popularity. From their combination, new and interesting cooperative services could be generated. For instance, gamers endowed with Augmented Reality (AR) visors, connected as wireless nodes in an ad‐hoc network, can interact with each other while immersed in the game. To enable this vision, we discuss here a hybrid architecture enabling game play in ad‐hoc mode instead of the traditional client‐server setting. In our architecture, one of the player nodes also acts as the server of the game, whereas other backup server nodes are ready to become active servers in case of network disconnection, ie, due to low energy level of the currently active server. This allows to have a longer gaming session before a disconnection occurs due to energy exhaustion. In this context, the server election strategy with the aim of maximizing network lifetime is not so straightforward. To this end, we have analyzed this issue through a Mixed Integer Linear Programming (MILP) model considering static network topologies and both numerical and simulation‐based analysis shows that the backup servers solution fulfills its design objective. Contemplating for mobility, we present a simulation study employing actual packet routing protocols and discuss the tradeoffs that emerge.

The Interference-aware Drone Ad-hoc Relay Network Configuration problem

Abstract Drones are more and more important in emergency response scenarios, where they can be equipped with wireless access points and deployed to configure a recovery ad-hoc flying network, e.g., to interconnect users positioned in a disaster area. Communication between users is allowed by means of multi-hop paths including one or more relay drones. An interesting optimization problem asks for determining the position of a minimum number of drones in order to guarantee that all the required communication paths can be established. Wireless channels are used and interference plays an important role: we propose an interference model and we embed it in a Mixed Integer Linear Programming formulation. Preliminary computational results are presented, showing the proposed model is able to solve instances of medium size in order of minutes.

A Heuristic for Multi-attribute Vehicle Routing Problems in Express Freight Transportation

We consider a multi-attribute vehicle routing problem arising in a freight transportation company owning a fleet of heterogeneous trucks with different capacities, loading facilities and operational costs. The company receives short- and medium-haul transportation orders consisting of pick-up and delivery with soft or hard time windows falling in the same day or in two consecutive days. Vehicle routes are planned on a daily basis taking into account constraints and preferences on capacities, maximum duration, number of consecutive driving hours and compulsory drivers rest periods, route termination points, order aggregation. The objective is to maximize the difference between the revenue from satisfied orders and the operational costs. We propose a two-levels local search heuristic: at the first level, a variable neighborhood stochastic tabu search determines the order-to-vehicle assignment, the second level deals with intra-route optimization. The algorithm provides the core of a decision support tool used at the planning and operational stages, and computational results validated on the field attest for an estimated 9% profit improvement with respect to the current policy based on human expertise.