An algorithm for sink positioning in bus-assisted smart city sensing

Highlights

• The communication delay of a mobile WSN is affected by sink positioning.

• The sink positioning problem with delay minimization can be modeled as an ILP.

• Our positioning algorithm finds results close to the optimal, in polynomial time.

• We can reduce 84% of sinks with less than 10% growth in the maximum delay.

Abstract

Smart Cities use data obtained from different sensors to offer better services. Such data is usually sent to sinks, using the paradigm of Internet of Things. However, covering the whole city area with static sensors might be expensive. This paper addresses the utilization of the public transportation system as a mobility platform for sensor nodes and bus stops acting as sinks. This platform can improve the coverage of a sensor network and take advantage of opportunistic communication. Since this platform can be used for delay-constrained applications, we propose an algorithm to choose sensor sinks in the bus stops of a smart city that minimizes the maximum delay when delivering messages. We compare our solution with an optimal formulation and use real data, obtained in buses from Rio de Janeiro, as an input of our algorithm. Our experiments show that we can use approximately 16% of bus stops as sinks, having less than 10% of increase in the network maximum delay and no significant loss in the spatial coverage.

Introduction

Projections show that human population should reach 8.5 billion people by 2030, mainly in urban areas,¹ posing many challenges to the cities and directly affecting the well-being of citizens. Smart Cities are an umbrella of different technologies to answer to those challenges. A recurrent strategy is to gather information about the city and intelligently use it to improve the services offered to the citizens, or create new services [1]. There are possible applications in weather, surveillance, pollution monitoring, and others [2]. Naturally, gathering data about a whole city is a huge challenge, and an infrastructure based on the Internet of Things (IoT) [3] is an important tool. The IoT paradigm is often denoted as the enhancement of daily-life objects with sensing, actuation, and communication capabilities [4]. In many scenarios, wireless sensor networks (WSN) are part of the IoT infrastructure [5]. WSNs are usually composed of many sensing nodes that sense data about the environment and send this data to one or more sinks. The sink, then, sends data to a so-called task manager node, responsible for storing, processing and serving data to the users [6].

Including mobile nodes in the wireless sensing network is a way to enlarge the sensed area [7]. In the context of a city, it is an alternative to having a (potentially expensive) fixed sensor network covering the whole city. In that scenario, mobile sensors opportunistically deliver data to gateways or to sinks [8]. In this paper, we depart from the observation that different metropolises count on bus lines on their public transportation system. We therefore build upon the idea of having buses as mobile sensing nodes. In the considered application scenario, buses are elements of the IoT infrastructure. This setup enables applications in the smart city dimensions of smart mobility, smart environment, and smart living [9]. Then, the problem of where and when buses will deliver data to the Internet arises. We assume that there will be sinks located at bus stops to receive the data and deliver it to the place in the cloud where the data will be concentrated, and decisions made based on the information gathered, such as taking actions over the city infrastructure. Another advantage of this sensing bus network is that bus mobility is predictable; buses are already part of the public infrastructure, meaning that the mobile nodes come for free, and the additional cost of having a bus carrying a sensor is small.

One of the issues with the opportunistic communication performed by sensor nodes embedded in buses is the delay to deliver data. Sensors have to wait until the bus is close to a sink in order to deliver the gathered data. On the other hand, the type of application defines the freshness required for the information to be useful [10]. Therefore, in this paper we analyze the communication latency experienced in a citywide sensing bus network. We consider a mobile WSN where sensor nodes are on board buses and sinks are located on a subset of the bus stops. Thus, we model the delay when the number of sinks is constrained, as an Integer Linear Programming (ILP) problem, and also propose an approximate polynomial algorithm to locate sinks at some of the bus stops. The polynomial algorithm is executed using as input real data of all the buses and bus stops of Rio de Janeiro. Our results show that installing sinks in only 16% of bus stops can increase network maximum delay in less than 10% with no loss to the area covered.

The paper is organized as follows. Section 2 presents related work and positions our proposal within the literature of the field. Section 3 models the application scenario of our sensing bus network. In Section 4, we model the problem of positioning sinks in the network as an ILP problem. In Section 5 we propose an approximate algorithm to select bus stops to install sinks, minimizing the network maximum delay and compare its performance with the optimal solution. In Section 6 we run the algorithm for a real dataset. Section 7 concludes the work and points out future directions for our work.