Joao F. M. Sarubbi

Designing mobile content delivery networks for the internet of vehicles

Abstract Content delivery is a key functionality for developing the Internet of Vehicles. In such networks, vehicles act as sensors of the urban mobility by constantly exchanging messages with another vehicles, the cellular network, and also the infrastructure (roadside units). However, the task of delivering content in such dynamic network is far from trivial. In this work, we investigate the development of Content Delivery Networks (CDN) in the context of vehicular networks. Roadside units support the communication by replicating and delivering contents to vehicles within their range of coverage. Initially, we devise a strategy for measuring the performance of the content delivery in vehicular networks. Then, we use the proposed metric for designing a deployment strategy allowing us to identify the better locations for deploying the roadside units in order to properly support the dissemination of a variety of contents, each content requiring specific levels of performance. We compare our deployment strategy to the intuitive strategy of allocating roadside units at the densest locations of the road network. The results demonstrate that our strategy requires less roadside units than the baseline for non-massive deployments in order to achieve similar levels of performance, incurring in less costs when setting up the content-delivery infrastructure.

A genetic algorithm for deploying roadside units in VANETs

In this work we propose a genetic algorithm, Delta-GA, for solving the allocation of Roadside Units (RSUs) in a Vehicular Network. Our goal is to find the minimum set of RSUs in order to meet a Deployment Δ^ρ1_ρ2. The Deployment Δ^ρ1_ρ2 is a metric for specifying minimal communication guarantees from the infrastructure supporting the Vehicular Network. We compare Delta-GA to two baseline algorithms, Delta-g and Delta-r, to solve the Deployment Δ^ρ1_ρ2. Our results demonstrate that Delta-GA requires less Roadside Units in order to achieve the same deployment efficiency.

Non-Intrusive Planning the Roadside Infrastructure for Vehicular Networks

In this article, we describe a strategy for planning the roadside infrastructure for vehicular networks based on the global behavior of drivers. Instead of relying on the trajectories of all vehicles, our proposal relies on the migration ratios of vehicles between urban regions in order to infer the better locations for deploying the roadside units. By relying on the global behavior of drivers, our strategy does not incur in privacy concerns. Given a set of α available roadside units, our goal is to select those α-better locations for placing the roadside units in order to maximize the number of distinct vehicles experiencing at least one V2I contact opportunity. Our results demonstrate that full knowledge of the vehicle trajectories are not mandatory for achieving a close-to-optimal deployment performance when we intend to maximize the number of distinct vehicles experiencing (at least one) V2I contact opportunities.

Delta-r: A novel and more economic strategy for allocating the roadside infrastructure in vehicular networks with guaranteed levels of performance

In this work we propose Delta-r, a new greedy heuristic for solving the allocation of roadside units in order to meet a Δρ2ρ1-Deployment. The Δρ2ρ1-Deployment is a metric for specifying minimal levels of performance from the infrastructure supporting vehicular networks. As far as we are concerned, this is the first QoS-bounded deployment strategy considering both the contact probability, and the contact duration. We compare Delta-r to two baselines: DL allocates the roadside units at the densest locations of the road network, while Delta-g uses the absolute V2I contact time. Differently from Delta-r, our proposal evaluates the deployment performance when using the relative V2I contact time considering vehicles and locations of the road network. Our results demonstrate Delta-r requiring less roadside units to achieve the same performance of the infrastructure supporting the V2I communication.

A strategy for clustering students minimizing the number of bus stops for solving the school bus routing problem

In this work we tackle the bus stop selection step for the School Bus Routing Problem (SBRP). Our goal is to minimize the number of bus stops in order to assign all students to a bus stop respecting a home-to-bus-stop walking distance constraint. Our strategy creates a large number of possible bus stops points in a road network and uses a pseudo-random constructive heuristic algorithm to assign students to a bus stops. Our approach is tested on a real georeferenced data of a Brazilian city and is compared with a different methodology. Results demonstrate that the proposed approach is able to find good solutions for this optimization problem. Besides, the higher the number of possible points to install bus stops, the smaller is the number of bus stops required to attend all students.

Gamma Deployment: Designing the Communication Infrastructure in Vehicular Networks Assuring Guarantees on the V2I Inter-Contact Time

Gamma Deployment is a metric for evaluating the distribution of roadside units in vehicular networks in terms of two parameters: a) the inter-contact time between vehicles and the infrastructure, and, b) the share of vehicles that must respect the inter-contact time guarantees. We envision the use of the Gamma Deployment metric when the network designer intends to distribute check-points along the road network in order to collect and disseminate traffic informations through roadside units. Thus, the goal is to locate the roadside units such that ρ percent of vehicles meet roadside units in time intervals less than τ seconds. In this work, we formalize the Gamma Deployment metric by developing an Integer Linear Programming formulation (ILP). Since the ILP is not able to solve large instances, we also develop an heuristic for approximating the optimal solution. We present experiments considering a realistic mobility trace, and our results demonstrate the heuristic incurs in small deviations for small inter-contact time guarantees.

Optimization of the vehicle routing problem with demand responsive transport using the NSGA-II algorithm

Demand Responsive Transport (DRT) systems emerge as an alternative to deal with the problem of variable demand, or even unpredictable, occurring in conventional urban transport systems. It can be seen in some practical situations such as public transport in rural areas, wherein in some situations, there is no way to predict demand. This paper addresses the Vehicle Routing Problem with Demand Responsive Transport (VRPDRT), a type of transport which enables customers to be taken to their destination like a taxi or minibus in order to reduce operating costs and to meet customer needs. A multiobjective approach is proposed to VRPDRT in which five different objective functions are used. These functions are aggregated in three new functions resulting in a three-objective formulation for VRPDRT. When using a three objective approach, that formulation allows a better understanding of the company and human perspectives while permitting to solve the resulting problem in an efficient way. The proposed three-objective optimization problem is solved using a random method of generating solutions and an algorithm considered state of the art, the Non-dominated Sorting Genetic Algorithm II (NSGA-II). The sets of solutions are compared using the Set Coverage Metric. The results show that the NSGA-II algorithm could obtain sets of solutions with better values for all objective functions used also called the non-dominated solutions set.

Multiobjective approach to the vehicle routing problem with demand responsive transport

The Vehicle Routing Problem (VRP) has been largely studied over the last years, since problems involving the transport of persons and/or goods have great practical application. This paper addresses the Vehicles Routing Problem with Demand Responsive Transport (VRPDRT), a type of transport which enables customers to be taken to your destination like a taxi or minibus in order to reduce operating costs and to meet customer needs. A multiobjective approach is proposed to VRPDRT in which five different objective functions are used. Using an iterative methodology, known as aggregation tree, the objective functions are used to construct a bi-objective version for the problem. The proposed bi-objective optimization problem is solved via NSGA-II and SPEA2 and the algorithm performances are compared using S-Metric. Through a statistical test, the results shows with 95% of confidence that the NSGA-II presents better convergence when compared with SPEA2.

A Cut-and-Branch algorithm for the Multicommodity Traveling Salesman Problem

This paper presents a Cut-and-Branch algorithm for the Multicommodity Traveling Salesman Problem (MTSP), a useful variant of the Traveling Salesman Problem (TSP). The MTSP presents a more general cost structure, allowing for solutions that consider the quality of service to the customers, delivery priorities and delivery risk, among other possible objectives. In the MTSP the salesman pays the traditional TSP fixed cost for each arc visited, plus a variable cost for each of the commodities being transported across the network. We present a strong mathematical formulation for this relevant problem. We implement a Cut-and-Branch algorithm for the MTSP which is able to find optimal solutions faster than stand-alone CPLEX codes.

Dimensionality Reduction Approach for Many-Objective Vehicle Routing Problem with Demand Responsive Transport

Demand Responsive Transport DRT systems emanate as a substitute to face the problem of volatile, or even inconstant, demand, occurring in popular urban transport systems. This paper is focused in the Vehicle Routing Problem with Demand Responsive Transport VRPDRT, a type of transport which enables passengers to be taken to their destination, as a shared service, trying to minimize the company costs and offer a quality service taking passengers on their needs. A many-objective approach is applied in VRPDRT in which seven different objective functions are used. To solve the problem through traditional multi-objective algorithms, the work proposes the usage of cluster analysis to perform the dimensionaly reduction task. The seven functions are then aggregated resulting in a bi-objective formulation and the algorithms NSGA-II and SPEA 2 are used to solve the problem. The results show that the algorithms achieve statistically different results and NSGA-II reaches a greater number of non-dominated solutions when compared to SPEA 2. Furthermore, the results are compared to an approach proposed in literature that uses another way to reduce the dimensionality of the problem in a two-objective formulation and the cluster analysis procedure is proven to be a competitive methodology in that problem. It is possbile to say that the behavior of the algorithm is modified by the way the dimensionality reduction of the problem is made.