sensors Article

An Intelligent Parking Management System for Urban Areas Juan A. Vera-Gómez 1 , Alexis Quesada-Arencibia 2, *, Carmelo R. García 2 , Raúl Suárez Moreno 1 and Fernando Guerra Hernández 1 1 2

*

Edosoft Factory, S.L. Las Palmas de Gran Canaria, Las Palmas de Gran Canaria 35011, Spain; [email protected] (J.A.V.-G.); [email protected] (R.S.M.); [email protected] (F.G.H.) Instituto Universitario de Ciencias y Tecnologías Cibernéticas, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria 35017, Spain; [email protected] Correspondence: [email protected]; Tel.: +34-928-457-108

Academic Editor: Vittorio M. N. Passaro Received: 28 April 2016; Accepted: 15 June 2016; Published: 21 June 2016

Abstract: In this article we describe a low-cost, minimally-intrusive system for the efficient management of parking spaces on both public roads and controlled zones. This system is based on wireless networks of photoelectric sensors that are deployed on the access roads into and out of these areas. The sensors detect the passage of vehicles on these roads and communicate this information to a data centre, thus making it possible to know the number of vehicles in the controlled zone and the occupancy levels in real-time. This information may be communicated to drivers to facilitate their search for a parking space and to authorities so that they may take steps to control traffic when congestion is detected. Keywords: intelligent transport systems; parking; wireless sensor network (WSN); traffic; sensor; infrared

1. Introduction Human mobility is a necessity in today’s world. It has a significant impact on both quality of life and the economy of modern societies. Transport systems are, therefore, a key element in developed or developing countries. Zhang [1], in his survey of Intelligent Transportation Systems, indicated that 40% of the world’s population spend at least one hour on the road every day. Of all the different modes of transport, the one that is used on a massive scale is land transport by road. Such large-scale use of this type of transport has led to congestion problems in densely populated metropolitan areas, with all the concomitant negative consequences. According to Shawe-Taylor [2], these negative consequences include pollution and its harmful effects on the environment and human health. Too much time on the road means an increase in energy consumption, which has a negative impact on both individual and national economies, as well as on the environment. Several medical studies [3] have confirmed that road transport congestion results in a deterioration of public health because it increases the risk of heart and respiratory diseases. Moreover, according to the WHO, over 7 million people die every year from health problems caused by pollution. One of the causes of this excessive amount of time spent on the road in private road transport is the need to spend time looking for free parking spaces. Pineda [4] studied the costs generated by the extra distance vehicles have to travel to find a parking space in the cities of Madrid and Barcelona. The costs in consumption for the extra distance and time spent on the road are approximately 347 million and 268 million euros per year, respectively. Public transport authorities and the operators of parking spaces are evaluating various solutions to improve parking space management. Solutions based on infrastructure investment are expensive and implementation is slow. Technology-based solutions have been proposed as an alternative with lower costs and faster implementation. Sensors 2016, 16, 931; doi:10.3390/s16060931

www.mdpi.com/journal/sensors

Sensors 2016, 16, 931

2 of 16

In this article we describe a low-cost, minimally-intrusive system for the efficient management of parking spaces on both public roads and controlled zones. This system is based on wireless networks of photoelectric sensors that are deployed on the access roads into and out of these areas. The sensors detect the passage of vehicles on these roads and communicate this information to a data centre, thus making it possible to know the number of vehicles in the controlled zone and the occupancy levels in real-time. This information may be communicated to drivers to facilitate their search for a parking space and to authorities so that they may take steps to control traffic when congestion is detected. The article is structured as follows: a selection of related studies is described in the second section, to provide context for our proposed system. The third section describes the system, explaining its general architecture, its main constituent elements, and the method developed to detect the passage of vehicles. Tests to verify system operation are described in the fourth section and, finally, the main conclusions are presented in section five. 2. Related Studies Various authors have looked at developing sensor-based technological solutions to improve the use of parking spaces. According to Bagula [5], intelligent vehicle parking space management systems may be classified according to the type of sensor detection. He distinguishes the systems that only monitor the entry or exit of vehicles from the parking area from the systems that are able to detect whether each parking space is occupied or free. Systems belonging to the first type are easier to deploy and less expensive, appropriate for monitoring the occupancy levels of large outdoor parking areas. Systems belonging to the second type provide more useful and more detailed information to users and may be combined with positioning and guidance services to help locate the available spaces. This type of system is used in indoor parking spaces and is more complex and expensive than the entry and exit monitoring systems, as it requires that each parking space is equipped with sensors and a more sophisticated communications infrastructure. Various parking space management system proposals are described below. Tang [6] proposed a wireless sensor network deployed in indoor car parks that shows the occupancy status of each parking space. Motes (sensor nodes) equipped with acoustic and light sensors are located in each space, and periodically notify whether the space is occupied or available. Benson [7] also proposed a network-based wireless sensor system. A communication link is established by ZigBee and the electromagnetic sensors were developed specifically for this system. Lin [8] proposed a vision-based parking management system to manage an outdoor car park using cameras set up around the parking space, sending information, including real-time display, to the ITS centre database. A scientific solution based on a GPS-based vehicle navigation system and the past and current status of the car park was proposed by Pullola [9], who modelled the availability of a car park using the Poisson process. The author also proposed an intelligent algorithm which helps the driver choose the parking space with the highest probability of being vacant. Lee [10] proposed a combination of magnetic and ultrasonic sensors to control car parks. This system is based on a modified version of the min-max algorithm for detection of vehicles using magnetometers and an algorithm for ultrasonic sensors. Srikanth [11] proposed an intelligent parking management system, consisting of a wireless network that uses different types of sensors to detect the presence of a vehicle in every one of the parking spaces; moreover, the system informs users and guides them to the location of the available space. The network’s sensor nodes communicate by radio frequency. Yoo [12] described a system, called S3, which is deployed in school zones, which is designed to detect and register vehicles driving at excessive speeds or parked in prohibited zones. This system consists of a wireless sensor network that is divided into two subnetworks: one to detect vehicles parked in prohibited zones and the other to detect vehicles travelling at excessive speeds. The sensors used are Anisotropic Magneto-Resistive (AMR) magnetic sensors, and the wireless communication link is established by ZigBee. Magrini [13] proposed a vision sensors network to monitor available spaces in public car parks, using distributed network nodes to perform the required processing and analysis of images. Chen [14] proposed a

Sensors 2016, 16, 931

3 of 16

system for locating available spaces in indoor car parks and a guidance system to locate the available space. The architecture of this system is based on a wireless network of ultrasound sensors that detect the presence of a vehicle in each of the parking spaces. The status of each space is transmitted by a sensor node that sends this information via RFID to special routing nodes, which communicate with each other to relay the data packets sent by the sensor nodes to the control centre. The network has a tree topology. Assistance in locating the space is provided by LEDs that indicate the status of the parking space. In the context of public parking space management based on image sensors, Salvadori [15] described image-processing techniques, using threshold algorithms, to monitor parking spaces, achieving high-performance detection of their occupancy status. Alessanderlli [16] proposed the ScanTraffic architecture, which is used to develop a system to monitor traffic flow and parking spaces at the International Airport of Pisa; this system is also based on a wireless image sensor network. Liu [17] described the iParking system, which is designed to facilitate the location of available spaces in an indoor car park. The system receives data from sensors located inside the car park, which indicate the spaces that are free, and is able to guide the driver to a free space using a positioning system for indoor spaces that is based on smartphone capabilities. Gu [18] proposed a system for managing parking spaces on public roads. The system is called Street Parking System (SPS), and uses a three-axis magnetic sensor to detect vehicles and ZigBee technology for wireless communications, achieving a reliability rate close to 99% in vehicle detection. Reve [19] also proposed a similar system, based on a wireless sensor network and LED display system that indicates the available spaces to drivers at the car park access points. The sensors are infrared and communications are established by radio frequency (RF). In the context of the Smart City paradigm, Giuffrè [20] proposed an IPA (Intelligent Parking Assistant) conceptual architecture that aims to overcome current parking management problems. Yang [21] presented a prototype Smart Parking Services system, based on Wireless Sensor Networks (WSNs), for finding free parking spaces. The proposed scheme consists of wireless sensor networks, an embedded web server, a central web server, and a mobile phone application. Each parking space has a light sensor node that detects the status of the parking space, reporting periodically to the embedded web server via the wireless sensor networks. Using a Wi-Fi network, this information is sent to the central web server in real time, displaying the status of the parking spaces on the driver’s mobile device. Geng [22] proposed a smart parking system for urban areas. The system’s functions include parking detection, reservation guarantee and Vehicle-to-Infrastructure (V2I) or Infrastructure-to-Vehicle (I2V) communication. Considering the requirements of the user, the system assigns and reserves an optimal parking space combining proximity to destination and parking cost, ensuring that the overall parking capacity is efficiently utilised. The system was tested in a garage of Boston University. Tian [23] proposed an intelligent parking management system based on License Plate Recognition (LPR), which recognises the licence plate automatically at the car park access point and provides vehicle information; experimental results show that this parking management system can achieve 95% accuracy, and can be applied to real-time implementation. Karunamoorthy [24] proposed an intelligent parking system that uses image-processing techniques to solve the problem of unnecessary time consumption in finding a parking space in commercial car parks. This parking management system provides information about the available parking spaces, as well as an automated payment system for registered users. Caballero-Gil [25] proposed a low-cost service to predict and manage indoor parking spaces. This service is based on a central system that predicts the available spaces in the car park using cellular automata and an application for smartphones that uses various technologies to help drivers find free parking spaces. In summary, this review of related studies shows that most of the proposed systems are designed to be deployed in indoor car parks and to monitor individual parking spaces, and that they usually include guidance services to locate the available space. From the point of view of the type of sensors used, these systems mainly use infrared and electromagnetic sensors. As for wireless communications, they use radio frequency, with ZigBee technology being the most used. In terms of network topology the most widely used is a tree topology.

Sensors 2016, 16, 931

4 of 16

The system proposed in this article is designed to monitor the entry and exit of vehicles into and out of outdoor parking zones located on public roads, not to monitor every single car space available, thus reducing the costs related to the deployment of technology. The system has been tested in closed car parks. On public roads, the system would not account for every single slot, but it is designed to provide estimations on real-time occupation levels. From a technological point of view, it uses a wireless network of photoelectric sensors, and uses 6LoWPAN over IEEE 802.15.4 for communications. It can be used across multiple communications platforms, using the IPv6 stack, which is hugely important to enable the IoT. IPv6 provides a basic transport mechanism to produce complex control systems and to communicate with other devices in a cost-effective manner using low-power wireless network. The network topology adopted for this system is a tree topology. As a result of mass production, the sensors used for the system have a very low cost in comparison with other sensors used for parking management systems, such as electromagnetic or pressure sensors. Despite their low cost, these sensors perform very well in detection applications; indeed, infrared sensors are widely employed in industrial automation processes, having proven reliability. Furthermore, their compact size makes it possible to enclose the whole node in a small box. In addition to this advantage, both the sensors themselves and the controlling boards have very low power consumption, and for this reason each sensor node is to be powered by a combination of batteries and solar power, minimising the deployment efforts and costs, as no wiring is needed. Another advantage of this approach over others is that it does not rely on the user’s involvement for the system to meet its purpose. There is no need for the vehicles to be equipped with smart location systems, or for the users to employ technological gadgets to access or update information about car park availability. The system is non-intrusive too: it does not identify the vehicles or their drivers at all, nor does it have to record or take pictures of them. Finally, this solution does not need more data processing than that done by the mainboards, further reducing infrastructure and system costs. Although it has not yet been developed, the system is intended to share parking availability data with the public authorities, contributing to a reduction in both traffic and its carbon footprint. In conclusion, the system, as will be seen in this article, is characterised by its ease of deployment, flexibility, and affordability when incorporating new nodes to monitor the access roads to the parking areas, and even new sensor subnets to monitor new parking areas. 3. Description of the System To monitor the number of available spaces in the parking zones a low-cost, low-energy WSN will be employed. This means that a complex roll-out or new cabling infrastructure will not be required in the zone where the proposed system is deployed. The sensors will detect the passage of vehicles on the roads into and out of the parking areas located on public roads in real-time. The authorities and drivers will, therefore, know the occupancy levels of the parking zones at all times. The real-time response is essential because the system would not be practicable if there was a considerable delay in updating these data. Relaying pseudo real-time information to drivers avoids them having to spend more time on the road searching for a parking space, which hinders the circulation of other vehicles and increases fuel consumption, emissions, and noise pollution. Therefore, the proposed system will enable transport regulators to manage public parking zones sustainably. In addition, to facilitate a scalable and flexible roll-out in large urban areas, the system can make use of virtualisation and computing paradigms to obtain the processing and storage resources required by the system architecture. This will enable comprehensive, centralised handling of the data provided by the sensor network in a control centre. Figure 1 shows the system being deployed on a public road and in a controlled parking area. The scheme represents the sensors and their direction on the access points and their communication with the gateway that relays the signal to the data centre. This gateway system may also provide in situ information on the occupancy level of both areas through an LED system.

Sensors 2016, 16, 931 Sensors 2016, 16, 931

5 of 16 5 of 16

Figure Figure 1. 1. System System architecture architecture in in aa closed closed car car park park and on a public road.

3.1. System System Architecture Architecture The The proposal proposal consists consists of of aa network network of of wireless wireless sensors sensors located located on on roads roads in and out of public parking areas. areas. These Thesesensors sensorsdetect detectthe theentry entryand and exit of vehicles into and of the parking areas, exit of vehicles into and outout of the parking areas, and and transmit this information vianetwork. the network. The sensors are connected to network nodes that transmit this information via the The sensors are connected to network nodes that receive receive data provided by the sensors and transmit themthe across the network; type of node isa the datathe provided by the sensors and transmit them across network; this typethis of node is called called a sensor node. In the network topology there is a node that enables the exchange of sensor node. In the network topology there is a node that enables the exchange of data betweendata the between the sensor and thethis data centre; this anode is called a Data gateway node.from Datathe obtained sensor nodes and thenodes data centre; node is called gateway node. obtained sensor from theare sensor network are stored in the datawhere centre, is also where the network stored and managed in theand datamanaged centre, which is also thewhich operation of each sensor operation of each sensor node is controlled. The data obtained by the data centre enable node is controlled. The data obtained by the data centre enable the provision of services that reportthe on provision of services report on the occupancy levels of the parking spaces on public roads. the occupancy levels that of the parking spaces on public roads. Each Each sensor sensor node node on on the network network will will be developed developed on a proprietary platform platform that that will will be responsible responsible for for collecting collecting the the data data obtained obtained by by the the sensor sensor or or sensors sensors associated associated with with that that node, node, interpreting network. The interpreting and packaging packaging information information and transmitting it over the network. The sensor sensor nodes must have havelow lowconsumption consumption levels, although when placed in outdoor parking it possible will be levels, although when placed in outdoor parking areas itareas will be possible addphotovoltaic a small photovoltaic to add a to small panel. panel. The gateway gateway node links the sensor nodes and the data centre. The sensor nodes notify the data centre when whena vehicle a vehicle enters or the exits the monitored parking area and receives orders that enters or exits monitored parking area and receives orders that development development or engineers maintenance engineers wish the network thevia data via the or maintenance wish to transmit to to thetransmit networktofrom the data from centre, thecentre, gateway. gateway. The data centre controls the sensor network and manages the data provided by it. To carry out data centre the sensor network andhas manages the data provided bynetwork it. To carry out theseThe procedures, thiscontrols component of the architecture two subsystems: the sensor control these procedures, this component of the architecture two subsystems: the sensor control subsystem that allows technical staff to configure andhas supervise the operations of thenetwork sensor network; subsystem that allows technicaldata staff to configure and supervise the operations of the sensor and the subsystem for recording provided by the sensors, which stores and processes the data in network; and theinformation subsystem for recording data provided by sensors, which processes order to provide services to users (authorities andthe drivers) about the stores status and of the parking the data in order by to provide information services to users about the statusand of areas monitored the system. To achieve scalability and(authorities flexibility inand the drivers) data centre computing the parking areas monitored by the system. To achieve scalability and flexibility in the data centre storage resources, we propose using cloud virtualisation and computing paradigms in the system. computing and storage resources, we propose using cloud virtualisation and computing paradigms in the system.

Sensors 2016, 16, 931

6 of 16

3.1.1. Sensor Nodes The sensor nodes send data packets that notify the entry or exit of a vehicle into or out of the monitored parking area. These packets include the identification number of the network6 ofand of the Sensors 2016, 16, 931 16 node to identify the parking area in question. Each sensor node interprets the messages received from Sensor Nodes the data3.1.1. centre. These packets contain the same fields as those sent by the node, but the identification fields of theThe network and in this toorthe terminal or terminals which the sensor nodesnode send will data correspond packets that notify thecase entry exit of a vehicle into or out to of the parking area.data These packets of thecentre network and sensor of the nodes. messagemonitored is directed, and the field will include containthe theidentification order sent number by the data to the to identify components the parking area in question. Each sensor node were interprets thea messages received Fornode the hardware of the nodes two circuit boards used: mainboard and a sensor from the data centre. These packets contain the same fields as those sent by the node, but the board. The mainboard includes the components responsible for data processing and communication identification fields of the network and node will correspond in this case to the terminal or terminals between nodes, while the sensor board is connected to the sensor. These two circuit boards, the batteries to which the message is directed, and the data field will contain the order sent by the data centre to and a solar panel are housed in a box with IP67 protection measuring 82 mm ˆ 80 mm ˆ 85 mm. the sensor nodes. The batteriesFor arethe lithium polymer 1900 mAh, with atwo nominal voltagewere of 3.7 V, and the solarand panel hardware components of the nodes circuit boards used: a mainboard a has a sensor board. The power output of 0.4 W (5 mainboard V/80 mA).includes the components responsible for data processing and between nodes, while the is sensor board is connected to the sensor. These two circuit Thecommunication main component of the mainboard the CC2530 system-on-chip from Texas Instruments [26], boards, the batteries and a solar panel are housed in a box with IP67 protection measuring 82 mm × a system that includes a microcontroller and a radio module in the integrated circuit. The 80 mm × 85 mm. The batteries are lithium polymer 1900 mAh, with a nominal voltage of 3.7 V, and microcontroller is the 16-bit Intel 8051 and has 8 KB of RAM and a programmable flash memory the solar panel has a power output of 0.4 W (5 V/80 mA). of 256 KB. The radio module is transceiver that in the 2.4 GHz is compatible The main component of athe mainboard is the operates CC2530 system-on-chip fromband Texas and Instruments with the[26], IEEE 802.15.4 programmable output power of up to 4.5 circuit. dBm, allowing a a system thatprotocol, includes a with microcontroller and a radio module in the integrated The microcontroller is the 16-bit of Intel 8051 and has 8 KB of component RAM and a programmable flash memory ofinto the maximum transmission speed 250 Kbps. Another that has been integrated 256 KB. TheCC2591 radio module is a transceiver that operates the 2.4Instruments, GHz band andwhich is compatible with mainboard is the amplifier, also manufactured byinTexas is compatible with the IEEE 802.15.4 protocol, with programmable output power of up to 4.5 dBm, allowing a the proposed radio system because it also operates in the 2.4 GHz band. This amplifier extends the maximum transmission speed of 250 Kbps. Another component that has been integrated into the range ofmainboard communications by providing a power amplifier output power and improves is the CC2591 amplifier, also manufactured by that Texasincreases Instruments, which is compatible receiverwith sensitivity. the proposed radio system because it also operates in the 2.4 GHz band. This amplifier extends theofrange communications by providing a power amplifier that increases output One the of main characteristics of the mainboard is its small size (50 ˆ 70power mm).andWireless improves receiver sensitivity. communication is established using the 6LoWPAN standard that enables the use of IPv6 over the IEEE One of the main characteristics of the mainboard is its small size (50 × 70 mm). Wireless 802.15.4 standard. The main objective of using this standard is to be able to use IPv6 on a low-energy communication is established using the 6LoWPAN standard that enables the use of IPv6 over the WSN, thereby enabling small devices, with limited processing power, to achieve broad connectivity IEEE 802.15.4 standard. The main objective of using this standard is to be able to use IPv6 on a (interoperability) maximise the battery life (Mulligan [27]). low-energy and WSN,tothereby enabling small devices, with limited processing power, to achieve broad Theconnectivity mainboard also has an and RS-232 interface, JTAG,life and a number (interoperability) to maximise the battery (Mulligan [27]). of expansion connectors The mainboard with also has an RS-232 interface, JTAG, and a number of expansion connectors allowing communication external devices. communication with external devices. As allowing shown in Figure 2, in addition to these devices, a 3 V DC/DC converter has been added to the As shown in Figure 2, in addition to these devices, a 3 V DC/DC converter has been added to mainboard to adapt the voltage supplied by the battery to that required by the integrated circuits. This the mainboard to adapt the voltage supplied by the battery to that required by the integrated is necessary because the nominal battery voltagebattery is 3.7voltage V. circuits. This is necessary because the nominal is 3.7 V.

Figure 2. Block diagram of the mainboard.

Figure 2. Block diagram of the mainboard.

Sensors 2016, 16, 931

7 of 16

Sensors 2016, 16, 931 Sensors 2016, 16, 931

7 of 16 7 of 16

As 3, the sensor boardboard is composed of a series interfaces and voltage converters As shown shownininFigure Figure 3, the sensor is composed of aofseries of interfaces and voltage As shown in Figure 3, the sensor board is composed of a series of interfaces and voltage to allow connection to a wide range of sensors. It has two DC/DC voltage converters, 12 V and 5 V, converters to allow connection to a wide range of sensors. It has two DC/DC voltage converters, 12 V converters to SDI-12, allow connection to amA wide range of sensors. It has two DC/DC voltage converters, 12 V an and a 4–20 mA interface. This circuit board also features a also battery charging systemcharging through andSDI-12, 5 V, an and a 4–20 interface. This circuit board features a battery and 5 V, ansolar SDI-12, and a solar 4–20 mA interface. This circuit board alsooffeatures a battery charging an external panel, thus increasing the autonomy ofthe the autonomy nodes. system through an external panel, thus increasing the nodes. system through an external solar panel, thus increasing the autonomy of the nodes.

Figure 3. Block diagram of the sensor board. Figure 3. Block diagram of the sensor board. Figure 3. Block diagram of the sensor board.

The photoelectric sensor used is a SHARP-GP2Y0D02YK0F (Sharp Corporation, Osaka, Japan) The photoelectric sensor a SHARP-GP2Y0D02YK0F (Sharp Corporation, Osaka, Japan) [28], photoelectric sensorused usedis is a SHARP-GP2Y0D02YK0F (Sharp Corporation, Osaka, Japan) [28], shown in Figure 4. shown in Figure 4. 4. [28], shown in Figure

Figure 4. SHARP-GP2Y0D02YK0F sensor alongside a standard AA battery to give an idea of its size. Figure idea of of its its size. size. Figure 4. 4. SHARP-GP2Y0D02YK0F SHARP-GP2Y0D02YK0F sensor sensor alongside alongside aa standard standard AA AA battery battery to to give give an an idea

This photoelectric sensor allows detection by the three types of standard sensing modes listed This1.photoelectric sensor allows detection by the three types of standard sensing modes listed in Table This photoelectric sensor allows detection by the three types of standard sensing modes listed in in Table 1. Table 1. Table 1. Standard sensing modes of the sensor. The chosen sensing mode is the diffuse mode, simple installation requirements, Table 1. Standard sensingjustified modes ofby theits sensor. with no calibration needed, and its detection performance. process, sensors Sensing Mode Advantages During the experimental Disadvantages Sensing Mode bumps. However, future Advantages Disadvantages were deployed in speed deployments could altered makeon use - be The range and depends theof - The range depends onmost the street furniture like bollards, street lights, or traffic lights. When those are not available, the characteristics of the of the straightforward installation is to deploy sensors on the ground, avoiding characteristics the need for retro-reflective object (colour, reflectivity, DIFFUSE object (colour, reflectivity, or through-beam sensors. - Ideal for short-range etc.). DIFFUSE - Ideal for short-range etc.). To count vehicles entering and leaving the car park, two photoelectric used to avoid applications. - Asensors highlywere reflective applications. A highly reflective false positives. If a person or an object othernot than a vehicle interrupts the sensor IR beam, it could - Does need a reflector. background may cause Does not need a reflector. background be incorrectly counted, but using two-- sensors and debugged controlling software, themay count will not Easy installation/alignment. false triggering ofcause the Easy installation/alignment. false triggering of the increase unless the pair of sensors actually detect a vehicle. This pair ofsensor. sensors is fitted on speed bumps located on the entry and exit roads to the parking zones, as shown in Figure 5. - sensor. Relatively short sensing - Relatively short sensing range. range. RETROREFLECTIVE Moderate sensing range. Shorter sensing range RETROREFLECTIVE -- Moderate sensing range. - Shorter sensing range than through-beam. Easy to align. -- Easy to align. - than May through-beam. detect reflections Requires assembly and May detect reflections - Requires assembly and wiring of only one from shiny objects. wiring of only one from shiny objects. emitter/receiver unit. - Requires reflector. emitter/receiver unit. - Requires reflector.

Figure 4. SHARP-GP2Y0D02YK0F sensor alongside a standard AA battery to give an idea of its size. Figure 4. SHARP-GP2Y0D02YK0F sensor alongside a standard AA battery to give an idea of its size.

This photoelectric sensor allows detection by the three types of standard sensing modes listed in Table 1. This photoelectric sensor allows detection by the three types of standard sensing modes listed Sensors 2016, 8 of 16 in Table 1.16, 931 Table 1. Standard sensing modes of the sensor.

Sensing Mode

Table 1. Standard sensing modes of the sensor. Table 1. Standard Advantages sensing modes of the sensor.

Disadvantages The range depends on the Sensing Mode Advantages Disadvantages characteristics of the Sensing Mode Advantages - TheDisadvantages range depends on the object (colour, reflectivity, characteristics of the The range depends on the Sensors 2016, 16,DIFFUSE 931 8 of 16 - Ideal for short-range etc.). object (colour,ofreflectivity, characteristics the object - A(colour, highly reflective etc.). DIFFUSE DIFFUSE - - applications. Ideal for Ideal for short-range etc.). reflectivity, - Requires proper - Does not need a reflector. background may cause short-range applications. - - A applications. A highly highlyreflective reflective alignment. - High margin for THROUGH-Beam background mayofcause -- Easy Doesinstallation/alignment. not need a reflector. false triggering the false - Does not needenvironments. a reflector. background may cause - Not recommended for triggering of the sensor. - contaminated Easy installation/alignment. sensor. Easy installation/alignment. false triggering of the detection transparent - Longer sensing range than -- Relatively Relatively of short short sensing sensor. range. sensing objects. other technologies. range. Relatively shorttosensing -- Requires space mount - Most reliable sensing mode RETROREFLECTIVE range. RETROREFLECTIVE and wire the emitter and for highlysensing reflective objects. -- Shorter sensing rangethan range. -- Moderate Moderate sensing range. Shorter sensing range RETROREFLECTIVE receiver individually. -- Easy to align. than through-beam. through-beam. Easy to align. - Moderate sensing range. - Shorter sensing range -- May Maydetect detect reflections reflectionsfrom and -- Requires Requiresassembly assembly and - Easy to align. than through-beam. shiny objects. wiring of only one by its simple installation requirements, wiring mode, of onlyjustified one from shiny objects. SensorsThe 2016,chosen 16, 931 sensing mode is the -diffuse - Requires May detect reflections8 of 16 Requires assembly emitter/receiver unit.and reflector. with no calibration needed, and its detection performance. During the--experimental process, sensors emitter/receiver unit. Requires reflector. wiring of only one from shiny objects. were deployed in speed bumps. However, future deployments could alteredproper and make use of - be Requires emitter/receiver unit. - Requires reflector. street furniture like bollards, street lights, or traffic lights. When those are not available, the most Requires proper alignment. alignment. High margin for THROUGH-Beam High margin for THROUGH-Beam Not recommended for straightforward installation is to deploy sensors on the ground, avoiding the need for retro-reflective contaminated environments. - Not recommended for contaminated environments. detection of or through-beam sensors. detection of transparent rangethan than -- Longer Longersensing sensing range transparent objects. othertechnologies. technologies. To count vehicles entering and leaving the car park, two photoelectric sensors were used to objects. other Requires space to mount and Mostreliable reliable sensing mode avoid false positives. If a person or--anMost object other sensing than a vehicle theemitter sensorto IRmount beam, it - Requires space mode interrupts wire the and for highly reflective objects. could be incorrectly counted, but using sensors and debugged theand count receiver individually. and wiresoftware, the emitter fortwo highly reflective objects. controlling will not increase unless the pair of sensors actually detect a vehicle. Thisreceiver pair of individually. sensors is fitted on speed bumps located on the entry and exit roads to the parking zones, as shown in Figure 5. The chosen sensing mode is the diffuse mode, justified by its simple installation requirements, with no calibration needed, and its detection performance. During the experimental process, sensors were deployed in speed bumps. However, future deployments could be altered and make use of street furniture like bollards, street lights, or traffic lights. When those are not available, the most straightforward installation is to deploy sensors on the ground, avoiding the need for retro-reflective or through-beam sensors. To count vehicles entering and leaving the car park, two photoelectric sensors were used to avoid false positives. If a person or an object other than a vehicle interrupts the sensor IR beam, it could be incorrectly counted, but using two sensors and debugged controlling software, the count will not increase unless the pair of sensors actually detect a vehicle. This pair of sensors is fitted on 5. located Installation ofthe the sensors in two speed bumps (left) photoelectric sensor at the speedFigure bumps on of the entry and roads to the parking zones, assensor shown in located Figure Figure 5. Installation sensors in exit two speed bumps (left) photoelectric located at the 5. centre centre of the bump; speed bump; of bumps speed bumps with the photoelectric sensors. of the speed (Right)(Right) Pair ofPair speed with the photoelectric sensors.

To develop the software for the sensor nodes we used the Contiki [29] operating system and the To develop the software for the sensor nodes we used the Contiki [29] operating system and the specific library for the CC2530 system-on-chip. Contiki is an open-source operating system geared specific library for the CC2530 system-on-chip. Contiki is an open-source operating system geared towards the Internet of Things. It enables communication with microcontrollers via the Internet and towards the Internet of Things. It enables communication with microcontrollers via the Internet and can run on a wide range of wireless low-power devices, providing mechanisms to estimate the total can run on a wide range of wireless low-power devices, providing mechanisms to estimate the total energy consumption of the system and to identify the units that consume the most. In addition, energy consumption of the system and to identify the units that consume the most. In addition, Contiki is designed for small systems with only a few kilobytes of memory available; it is very Contiki is designed for small systems with only a few kilobytes of memory available; it is very efficient efficient at memory management and provides different memory allocation mechanisms. With at memory management and provides different memory allocation mechanisms. With regard to regard to communications, Contiki is fully compatible with the IPv6 and IPv4 standards, providing a communications, Contiki is fully compatible with the IPv6 and IPv4 standards, providing a full IP full IP network stack, with standard suchbumps as User Datagram Protocol (UDP), Transmission Figure 5. Installation of the sensorsprotocols insuch two as speed (left) photoelectric sensor located at the network stack, with standard protocols User Datagram Protocol (UDP), Transmission Control Control Protocol (TCP), and Hypertext Protocol (HTTP). Furthermore, centre of the speed bump; (Right) Pair of Transfer speed bumps with the photoelectric sensors. it also supports Protocol (TCP), and Hypertext Transfer Protocol (HTTP). Furthermore, it also supports low-power low-power wireless communications on the 6LoWPAN, Routing Protocol for Low-Power and Lossy wireless communications on the 6LoWPAN, Routing Protocol for Low-Power and Lossy Networks Networks (RPL)the andsoftware CoAP standards. To develop for the sensor nodes we used the Contiki [29] operating system and the (RPL) and CoAP standards. specific library for the CC2530 system-on-chip. Contiki is an open-source operating system geared 3.1.2. Gateway towards the Internet of Things. It enables communication with microcontrollers via the Internet and can run a wide range of wireless providing mechanisms estimate Theongateway node receives the low-power packets sentdevices, by the sensor nodes and sendsto them on tothe thetotal data energy consumption of the system and to identify the units that consume the most. In addition, centre. It also links communications in the opposite direction. From the hardware point of view, this Contiki forofsmall systems with only fewused kilobytes of memory available; it is very elementisisdesigned composed a mainboard similar to athat for the sensor nodes. Therefore, the efficient at memory management and provides different memory allocation mechanisms. With

Sensors 2016, 16, 931

9 of 16

3.1.2. Gateway The gateway node receives the packets sent by the sensor nodes and sends them on to the data centre. It also links communications in the opposite direction. From the hardware point of view, this element is composed of a mainboard similar to that used for the sensor nodes. Therefore, the gateway software will have characteristics and functionalities similar to that used for the sensor nodes without having to perform data packing or unpacking. The gateway will communicate with the sensor nodes using the 6LoWPAN communication protocol. If there is a node in the data centre acting as a border router, communication between the base station and the gateway will be done by 6LoWPAN. However, if the gateway is located very close to the data centre, a RS-232 serial connection will be used. 3.1.3. Data Centre The data centre will receive information transmitted by the gateway and from the sensor nodes of the network, giving readings which indicate the passage of vehicles. It will also send the relevant instructions to the sensor network and store system data. A server component will be installed to read and send data to the wireless sensor network while, for storing system information, a registration component will be used in the data centre. The server component for reading and sending data to the sensor network was developed as a web service, using Play Framework with Java technology. This framework is inspired by Ruby on Rails and Django, and enables flexible web applications to be built. It is a modular framework with integrated unit testing (it includes support for JUnit and Selenium) that allows static methods, non-blocking asynchronous communication, and simple corrective maintenance. The registration component consists of a database which will store information on registered users, the sensor networks deployed, the number of parking spaces available, etc. Specifically, we used PostgreSQL to set up the database. PostgreSQL functions include transactions, referential integrity, views and a multitude of features. It also incorporates Multiversion Currency Control—MVCC—which allows other clients to access it while a process is being written in a table without the need for locking. These features are suitable for systems that need to handle large amounts of data and a high number of concurrent users accessing the system simultaneously, as will be required by the proposed system. 3.2. Communications System Communication between the sensor nodes and the gateway node will be performed using the 6LoWPAN protocol on the Contiki operating system. This protocol allows the use of IPv6 over wireless networks implemented using the IEEE 802.15.4 standard. In the proposed system each sensor node is assigned an IP address. The transmission of packets using the IPv6 protocol requires an adaptation layer between IPv6 and the IEEE 802.15.4 standard. One of the functions of this adaptation layer is to compress the IP header, as this header is too large for 802.15.4. It also handles the fragmentation and reconstruction of packets because IPv6 supports a maximum transfer unit of 1280 bytes, while the maximum for the IEEE 802.15.4 standard is 127 bytes. It also handles the routing of packets within the local IP network, routing to external IP networks, readings of the status of neighbouring nodes, and support for multicast transmission. In the IEEE 802.15.4 standard, node consumption during transmission varies between 10 and 30 mA, depending on the power level, and communication is performed with a maximum transfer rate of 250 kbit/s. IEEE 802.15.4 uses the ISM band for industrial, scientific, and medical uses: 868 MHz in Europe, 915 MHz in the US, and 2.4 GHz worldwide. However, when designing devices for wireless sensor networks, the 2.4 GHz band is usually chosen since it is not assigned. It should be noted that the sensor network will be implemented following the ad hoc topology. This network topology describes networks as flexible mesh. In these networks all of the nodes provide routing services, so, in addition to functioning as end devices, they also perform packet retransmission

Sensors 2016, 16, 931

10 of 16

multi-hop functions. This way it is possible to transmit packets to nodes with which there is no direct coverage. Thanks to this feature, the network can cover large areas. However, this topology requires routing protocols to calculate the most efficient route to favour lower energy consumption in the nodes. Figure 6 shows the payload of a 6LoWPAN packet structure used in the system for communicating with the nodes. The payload consists of a custom header, IDs for each specific sensor node, the data length and the data itself (a message determining whether a vehicle went in or out). These messages Sensors 2016, 16, 10 are sent using 6LoWPAN protocol under Contiki. Sensors 2016, 16, 931 931 10 of of 16 16

Figure Structure of Figure 6. 6. Structure of data data packet packet sent sent by by the the sensor sensor nodes. nodes.

3.3. 3.3. Detection Detection of of the the Passage Passage of of Vehicles Vehicles This describes the method used to detect the passage vehicle on road into out ofof a This section section describes describesthe themethod methodused usedto todetect detectthe thepassage passageofof ofaaavehicle vehicleon ona aaroad roadinto intooror orout out of aa parking area. The method is designed to detect cars, vans or larger vehicles of two or more axles. parking area. The method is designed to detect cars, vans or larger vehicles of two or more axles. parking area. The method is designed to detect cars, vans or larger vehicles of two or more axles. The The sensor sensor node node is is connected connected to to two two sensors sensors that that detect detect the the passage passage of of vehicles. vehicles. Figure Figure 77 shows shows the solution proposed using the GP2Y0D02YK0F SHARP (Sharp Corporation, Osaka, Japan) solution proposed using the GP2Y0D02YK0F SHARP (Sharp Corporation, Osaka, Japan) infrared the solution proposed using the GP2Y0D02YK0F SHARP (Sharp Corporation, Osaka, Japan) infrared sensor. Each the and S2 aa speed bump protection and sensor. of the twoof and S2 S1 was placed in aplaced speed in bump for protection as a deterrent infraredEach sensor. Each ofsensors the two twoS1sensors sensors S1 and S2 was was placed in speed bump for for and protection and as as ato to encourage vehicle’s wheels to side the bump. encourage vehicle’sthe wheels to pass on each sideon ofeach the speed The distance, L, distance, between a deterrent deterrent tothe encourage the vehicle’s wheels to pass pass on each side of ofbump. the speed speed bump. The The distance, L, the fixed to vehicles than aa certain The exit the two bumps wasbumps fixed inwas order to in detect vehicles longer thanlonger a certain Thelength. exit and entry L, between between the two two bumps was fixed in order order to detect detect vehicles longer thanlength. certain length. The exit and entry monitoring process was carried out by controlling the activation and deactivation event of monitoring process was carriedwas outcarried by controlling the activation deactivation event of each of the and entry monitoring process out by controlling theand activation and deactivation event of each of the sensors (On/Off). sensors (On/Off). each of the sensors (On/Off).

Figure 7. Figure 7. 7. Deployment Deployment of of the the sensors. sensors. Figure Deployment of the sensors.

Knowing Knowing the the precise precise situation situation of of the the two two sensors sensors and and the the distance distance between between them, them, this this Knowing allows the precise situation of athe two sensors and the distance between them, thistoconfiguration configuration the passage of vehicle and its direction to be detected, thanks the configuration allows the passage of a vehicle and its direction to be detected, thanks to the order order of of allows the and passage of a vehicle and its direction to be2). detected, to the order of activation and activation deactivation of sensors (see When car or the activation and deactivation of the the sensors (see Table Table 2). When aathanks car is is entering entering or exiting, exiting, the system system deactivation ofidle the state, sensors Table When aactive. car is Then, entering exiting,or system startsactivates from an starts which no sensor is an entering exiting vehicle starts from from an an idle state, in in (see which no 2). sensor is active. Then, an or entering orthe exiting vehicle activates idle state, in which noS2 sensor is active.As Then, an entering or exiting vehicle activates the first sensor, the sensor, S1 respectively. the continues its the sensor will the first first sensor, S1 or or S2 respectively. As the vehicle vehicle continues its movement, movement, the second second sensor will S1 or S2 respectively. As the vehicle continues its movement, the second sensor will be activated: S2 be to be activated: activated: S2 S2 for for entering entering vehicles vehicles and and S1 S1 for for exiting exiting vehicles. vehicles. In In the the final final stage stage before before returning returningfor to entering vehicles and S1 for exiting vehicles. In the final stage before returning to an idle state, only one an idle state, only one sensor would be active: S2 for entering vehicles and S1 for vehicles moving in an idle state, only one sensor would be active: S2 for entering vehicles and S1 for vehicles moving in sensor woulddirection. be active: S2 for entering vehicles and S1 for vehicles moving in the opposite direction. the the opposite opposite direction. Table Table 2. 2. Map Map of of direction direction vector vector control control events events with with two two in-line in-line sensors. sensors.

Entry Entry S1 S1 Off Off On On On On

Exit Exit S2 S2 Off Off Off Off On On

S2 S2 Off Off On On On On

S1 S1 Off Off Off Off On On

Sensors 2016, 16, 931

11 of 16

Table 2. Map of direction vector control events with two in-line sensors. Entry

Exit

S1

S2

S2

S1

Off On On Off Off

Off Off On On Off

Off On On Off Off

Off Off On On Off

Sensors 2016, 16, 931

11 of 16

However, this configuration of sensor pairs is not sufficient to avoid false positives. To decrease the error rate, two timers, T1 and T2, added toby the monitoring process, one fora each sensor, avoid false positives. False positives canwere be triggered a moving object; for instance, person with to avoid falsetrolley, positives. Falsethrough positives can be triggered object; for instance, a personorwith a shopping passing the sensor system,by oraifmoving a vehicle reverses at the entrance, if a avehicle shopping trolley, passing system, if a vehicle reverses at the entrance, or if a uses the lane to through perform the a sensor manoeuvre to or change direction. Therefore, the distance vehicle uses between the lane to perform manoeuvre to change direction. Therefore, the distance parameters the sensorsa and the timers monitoring the time of passage should parameters be set with between precision.the sensors and the timers monitoring the time of passage should be set with precision. The detection follows the the sequence diagram shownshown in Figure This diagram detectionalgorithm algorithm follows sequence diagram in 8. Figure 8. This describes diagram how the T1 and T2 values are associated with each of the sensors to detect different events: passage of describes how the T1 and T2 values are associated with each of the sensors to detect different events: apassage vehicle,ofvehicle reverses when passing through, orthrough, the vehicle a vehicle, vehicle reverses when passing or is thestationary. vehicle is stationary.

Figure 8. Vehicle entry detection detection sequence sequence diagram. diagram. Figure 8. Vehicle entry

As the diagram shows, it is necessary to check that the vehicle has been detected by the two As the diagram shows, it is necessary to check that the vehicle has been detected by the two pairs pairs of sensors. This avoids false positives in the event that the vehicle reverses and does not enter, of sensors. This avoids false positives in the event that the vehicle reverses and does not enter, or leaves or leaves the parking area. the parking area. As the vehicle enters, a reading is taken from both sensors every second. If sensor 1 is activated, As the vehicle enters, a reading is taken from both sensors every second. If sensor 1 is activated, timer T1 is initiated to verify that the sensor has been activated by a vehicle and not a person or other timer T1 is initiated to verify that the sensor has been activated by a vehicle and not a person or other object, and once the wait time has passed, a reading is taken from sensor 2. object, and once the wait time has passed, a reading is taken from sensor 2. If sensor 1 is still active but sensor 2 has not been activated, it means that the vehicle has If sensor 1 is still active but sensor 2 has not been activated, it means that the vehicle has stopped. stopped. If, however, sensor 1 is deactivated without sensor 2 having been activated the vehicle has If, however, sensor 1 is deactivated without sensor 2 having been activated the vehicle has reversed. reversed. If sensor 2 has been activated and sensor 1 remains active, timer T2 is initiated to verify that sensor 2 has been activated by a vehicle. If none of the sensors is then deactivated it may mean that the vehicle is stationary or that a long vehicle is passing. Lastly, a reading of the status of both sensors is taken, in the expectation that sensor 1 has been deactivated before sensor 2, which means that the vehicle has entered the parking area.

Sensors 2016, 16, 931

12 of 16

If sensor 2 has been activated and sensor 1 remains active, timer T2 is initiated to verify that sensor 2 has been activated by a vehicle. If none of the sensors is then deactivated it may mean that the vehicle is stationary or that a long vehicle is passing. Lastly, a reading of the status of both sensors is taken, in the expectation that sensor 1 has been deactivated before sensor 2, which means that the vehicle has entered the parking area. To detect a vehicle leaving the parking area, the same process is followed in reverse: first sensor 2 has to be activated and then sensor 1. In Figure 9 we can see the role that the timer plays in detecting vehicle entry; in this case Timer1 (T1), associated with Sensor1 (S1) has been activated after an Off signal from S1. Therefore, it is Sensors 2016, 16, 931 12 of 16 not counted as an attempted entry since the vehicle would have had to reach a speed of more than 20 km/h. Sensors 2016, 16, 931 12 of 16

Figure 9. Sequencediagram diagram showing showing aafailed to to enter. Figure 9. Sequence failedattempt attempt enter. Figure 9. Sequence diagram showing a failed attempt to enter. The “vehicle detection” event, both at the entry and exit control points, is notified by sending a The “vehicle detection” event, both at the the entry entry and exit control points, is Following notified by by sending data“vehicle frame from the sensorevent, node to the at gateway node, then on control to the data centre. thesending data The detection” both and exit points, is notified aa data frame from the sensor node to the gateway node, then on to the data centre. Following the data frame structure described in Section 3.2 and shown in Figure 6 for notification of entry or exit of a data frame from the sensor node to the gateway node, then on to the data centre. Following the data the data packet in sent by the sensor node will be as shown in Figure a header of of framevehicle, structure described in Section Section 3.2 and and shown instructured Figure 66 for for notification of10: entry or exit exit of aa frame structure described 3.2 shown in Figure notification of entry or six bytes with the values 53 hex , 45 hex , 4E hex ,53 hex , 4F hex , and 52 hex (the word SENSOR) followed by the vehicle, the the data datapacket packetsent sentby bythe thesensor sensornode nodewill willbebestructured structured shown Figure a header vehicle, asas shown inin Figure 10:10: a header of sensor node identifier, the53 sensor identifier, the,53 number of bytes in52 the data field of the frame, which of six bytes with the values , 45 , 4E , 4F , and (the word SENSOR) followed six bytes with the values 53hex, hex 45hex, 4E hex, 4F hex, and word SENSOR) followed by the hexhex ,53 hex hex hex52hex (thehex in this case is 1 and, finally, the data field containing the value 1hex if notifying vehicle entry or a 0hex by the node sensoridentifier, node identifier, the sensor identifier, the number ofinbytes in the data of the which frame, sensor the sensor identifier, the number of bytes the data field offield the frame, if notifying vehicle exit.

which thisiscase is 1finally, and, finally, the field data containing field containing the value if notifying vehicle in this in case 1 and, the data the value 1hex if1hex notifying vehicle entryentry or a or 0hexa 0 if notifying vehicle exit. ifhex notifying vehicle exit.

Figure 10. Structure of data packet sent by the sensor nodes to notify vehicle entry or exit.

4. Tests Figure 10. Structure of data packet sent by the sensor nodes to notify vehicle entry or exit. Figure 10. Structure of data packet sent by the sensor nodes to notify vehicle entry or exit. To test the proposed system under real conditions, it was installed in a parking area located on the Tafira Campus of the University of Las Palmas de Gran Canaria. This parking area has a capacity 4. Tests for 10 vehicles and has a single lane that is used both to enter and exit; the setup is shown in Figure tests were conducted verify the behaviour of installed the following of area the system: To test 11. theThe proposed system undertoreal conditions, it was in aaspects parking located on communications system; establishing the configuration parameters of the system the Tafira Campus of the University of Las Palmas de Gran Canaria. This parking areafor hascorrect a capacity detection of the passage of a vehicle (distance between the two speed bumps fitted with a sensor and

for 10 vehicles and has a single lane that is used both to enter and exit; the setup is shown in

Sensors 2016, 16, 931

13 of 16

4. Tests To test the proposed system under real conditions, it was installed in a parking area located on the Tafira Campus of the University of Las Palmas de Gran Canaria. This parking area has a capacity for 10 vehicles and has a single lane that is used both to enter and exit; the setup is shown in Figure 11. The tests were conducted to verify the behaviour of the following aspects of the system: communications system; establishing the configuration parameters of the system for correct detection of the passage of a vehicle (distance between the two speed bumps fitted with a sensor and the time each sensor indicating the presence of a vehicle must be active); and incorrect detection of the passage of a vehicle due to2016, a false positive. Sensors 16, 931 13 of 16

Figure 11. 11. Deployment of prototype prototype sensor sensor system system to to monitor monitor access access to to the the parking parking area. area. The The sensors sensors Figure Deployment of are installed in the speed bumps. are installed in the speed bumps.

To check the communications system was functioning correctly, an automatic periodic To check the communications system was functioning correctly, an automatic periodic verification verification process was developed to check the status of each sensor node. Specifically, the sensor process was developed to check the status of each sensor node. Specifically, the sensor nodes were nodes were programmed, once they had been activated from the data processing centre, to send the programmed, once they had been activated from the data processing centre, to send the status of their status of their batteries every hour over a period of seven days. It was found that these were batteries every hour over a period of seven days. It was found that these were correctly identified by correctly identified by their IPv6 address and that the frames were sent correctly without errors and their IPv6 address and that the frames were sent correctly without errors and correctly routed through correctly routed through the gateway node to the central data processing system. One hundred the gateway node to the central data processing system. One hundred sixty-eight packages of 168 were sixty-eight packages of 168 were received, with an error rate of 0%. The weather conditions during received, with an error rate of 0%. The weather conditions during the test were: clear sky, cloudy sky, the test were: clear sky, cloudy sky, and light rain. and light rain. Bearing in mind the way that the method of detecting passing vehicles was designed, two Bearing in mind the way that the method of detecting passing vehicles was designed, two values values need to be correctly configured for the process to run reliably: need to be correctly configured for the process to run reliably:  The distance separating the sensors placed in each speed bump. This is a fixed value that we ‚ The separating the sensors placed in each speed bump. This is a fixed value that we have havedistance called L. called L. that each sensor of the pair controlled by a node sensor needs to be active, thus  The time ‚ The time that sensorofofathe pair controlled byisa anode sensor be active, thus indicating indicating theeach presence moving object. This value that needs can beto varied by reprogramming the of a moving object. This is a value be varied by reprogramming sensor the presence sensor nodes. We have called these valuesthat T1 can (value of sensor 1 timer) and T2the (value of nodes. have T1 called these values T1 (value of sensor 1 timer) and T2 (value of sensor 2 timer); sensor We 2 timer); = T2. T1 = T2. The distance separating the two sensors, each installed in a speed bump, is a fixed value and, for aThe correct setting, the lengths of the different of vehicles on the market must value be taken distance separating the two sensors, eachtypes installed in a speed bump, is a fixed and,into for For this,the we conducted study oftypes vehicle lengthson (see 3),must concluding most aaccount. correct setting, lengths of theadifferent of vehicles theTable market be takenthat intothe account. appropriate distance isa study 2.6 m,of corresponding to the shortest wheelbase, longer than the For this, we conducted vehicle lengths (see Table 3), concluding that but the most appropriate wheelbase moving objects, as a bicycle or a motorcycle. distance is of 2.6other m, corresponding to such the shortest wheelbase, but longer than the wheelbase of other moving objects, such as a bicycle or a motorcycle. Table 3. Values of timers T1 and T2, depending on vehicle type, at a constant speed of 20 km/h.

Vehicle Type Bus Articulated bus Car Bicycle

Wheelbase 15 m 18.75 m Variable: 6.092 m max. 2.6 m min. 1.1 m

Value (s) 2.7 3.24 0.468–1.08 0.18

The time that each sensor must be active indicating the presence of a moving object, T1 for sensor 1 and T2 for sensor 2, depends on the speed of the vehicle in the parking area access lane; the

Sensors 2016, 16, 931

14 of 16

Table 3. Values of timers T1 and T2, depending on vehicle type, at a constant speed of 20 km/h. Vehicle Type

Wheelbase

Value (s)

Bus

15 m

2.7

Articulated bus

18.75 m

3.24

Car

Variable: 6.092 m max. 2.6 m min.

0.468–1.08

Bicycle

1.1 m

0.18

The time that each sensor must be active indicating the presence of a moving object, T1 for sensor 1 and T2 for sensor 2, depends on the speed of the vehicle in the parking area access lane; the maximum speed allowed in parking areas is 20 km/h. Table 3 shows these values for different types of vehicles travelling at 20 km/h. False positive situations are impossible to avoid, due to moving objects other than vehicles entering the parking zone, or an intentional improper use of the deployed system. However, these kinds of errors did not appear during the test period, although false negative values were detected. These were caused by vehicles, generally short vehicles, not passing over both sensors situated on the ground, but avoiding one of them when entering or leaving the parking zone. These tests were performed using conventional passenger cars, not other types of vehicles, like trucks or buses, due to parking size restrictions. 5. Conclusions One way to mitigate the negative effects of heavy vehicle traffic in urban and metropolitan areas is the effective management of vehicle parking spaces. Technology has been mooted as the basis for solutions that will have the fastest impact and lowest cost. This article has described a scalable, low-cost, and minimally-intrusive parking space monitoring system, using a wireless sensor network based on IPv6, using the 6LoWPAN protocol, a cloud-based management system and photoelectric sensors. The network sensors are located on speed bumps placed on the way in or out of the parking areas; these sensors are connected to nodes capable of processing the signal sent by the sensors to detect the entry or exit of vehicles into or out of the parking area. These nodes communicate wirelessly with in situ signalling devices and gateway units that send the signal to the central management server for processing or distribution via the Internet. The system architecture and the elements that configure it facilitate deployment in parking zones located on public roads. The number of detection monitoring units is low, and the system can, therefore, be scaled up to a city-wide level, monitoring entire areas, as long as all the entry and exit points are controlled. To check the system, it was tested under real conditions monitoring a car park on a university campus. The tests showed that the system works correctly with an error rate of less than 1%. For future research, we propose reducing installation impact by incorporating the solar power system and the processing and communication node together with the sensors in the speed bump, as this would further facilitate adoption of this monitoring technology. In addition, tests should be conducted in traffic zones with higher speeds—50 and 60 km/h—in order to be able to implement the system on fast lane exit points and, thus, monitor entire districts of a city. Acknowledgments: This Project has been funded by code TSI-020601-2012-47 under AETySI—Competitividad I+D programme of the Spanish Ministry of Industry, Energy and Tourism (sponsors). Thanks to Instituto Universitario de Ciencias y Tecnologías Cibernéticas of ULPGC for the right for use its parking area for prototype deployment. Thanks also to every member of Edosoft Factory Lab Team for their collaboration in the project. Author Contributions: Juan A. Vera Gómez directed the research study; all authors contributed to the design of the system architecture, the hardware selection and the tests; Alexis Quesada-Arencibia and Carmelo R. García developed the data centre software module and the communication system. Juan A. Vera Gómez, Raúl Suárez Moreno and Fernando Guerra Hernández developed the sensor nodes software module and the gateway software module; all authors contributed to the preparation of this manuscript.

Sensors 2016, 16, 931

15 of 16

Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations The following abbreviations are used in this manuscript: WHO WSN IPv6 IPv4 UDP TCP HTTP 6LoWPAN RPL CoAP MVCC

World Health Organization Wireless Sensor Network Internet Protocol version 6 Internet Protocol version 4 User Datagram Protocol Transmission Control Protocol Hypertext Transfer Protocol IPv6 over Low power Wireless Personal Area Networks IPv6 Routing Protocol for Low-Power and Lossy Networks Constrained Application Protocol Multiversion Concurrency Control

References 1.

2. 3. 4. 5. 6.

7.

8. 9.

10. 11.

12. 13. 14.

Zhang, J.; Wang, F.; Wang, K.; Lin, W.; Xu, X.; Chen, C. Data-Driven Intelligent Transportation Systems: A Survey. In IEEE Transactions on Intelligent Transportation Systems; IEEE: Hoboken, NJ, USA, 2011; Volume 12, pp. 1624–1639. Shawe-Taylor, J.; de Bie, T.; Cristianini, N. Data mining, data fusion and information management. In IEE Proceedings—Intelligent Transport Systems; IET: Hertfordshire, UK, 2006; Volume 153, pp. 221–229. Peters, A.; von Klot, S.; Heier, M.; Trentinaglia, I.; Hörmann, A.; Wichmann, H.E.; Löwel, H. Exposure to traffic and the onset of myocardial infarction. N. Engl. J. Med. 2004, 351, 1721–1730. [CrossRef] [PubMed] Pineda, M.; Abadia, X. Criterios de Movilidad, el estacionamiento urbano en superficie. Fund. RACC 2011, 1, 1–58. Bagula, A.; Castelli, L.; Zennaro, M. On the Design of Smart Parking Networks in the Smart Cities: An Optimal Sensor Placement Model. Sensors 2015, 15, 15443–15467. [CrossRef] [PubMed] Tang, V.W.S.; Zheng, Y.; Cao, J. An Intelligent Car Park Management System based on Wireless Sensor Networks. In Proceedings of the 1st International Symposium on Pervasive Computing and Applications, Urumqi, China, 3–5 August 2006; pp. 65–70. Benson, J.P.; O’Donovan, T.; O’Sullivan, P.; Roedig, U. Car-Park Management using Wireless Sensor Networks. In Proceedings of the 31st IEEE Conference on Local Computer Networks, Tampa, FL, USA, 14–16 November 2006; pp. 588–595. Lin, S.; Chen, Y.; Liu, S. A Vision-Based Parking Lot Management System. In Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, Taipei, Taiwan, 8–11 October 2006. Pullola, S.; Atrey, P.K.; el Saddik, A. Towards an Intelligent GPS-Based Vehicle Navigation System for Finding Street Parking Lots. In Proceedings of the IEEE International Conference on Signal Processing and Communications, Dubai, UAE, 24–27 November 2007; pp. 1251–1254. Lee, S. Intelligent parking lot application using wireless sensor networks. In Proceedings of the International Symposium on Collaborative Technologies and Systems, Irvine, CA, USA, 19–23 May 2008; pp. 48–57. Srikanth, S.V.; Pramod, P.J.; Dileep, K.P.; Tapas, S.; Patil, M.U.; Sarat Chandra Babu, N. Design and Implementation of a prototype Smart PARKing (SPARK) System using Wireless Sensor Networks. In Proceedings of the International Conference on Advanced Information Networking and Applications Workshops (WAINA ‘09), Bradford, UK, 26–29 May 2009; pp. 401–406. Yoo, S.; Chong, P.; Kim, D. S3: School Zone Safety System Based on Wireless Sensor Network. Sensors 2009, 9, 5968–5988. [CrossRef] [PubMed] Magrini, M.; Moroni, D.; Nastasi, C.; Pagano, P.; Petracca, M.; Pieri, G.; Salvadori, C.; Salvetti, O. Visual Sensor Networks for Infomobility. Pattern Recognit. Image Anal. 2011, 21, 20–29. [CrossRef] Chen, M.; Chang, T. A parking guidance and information system based on wireless sensor network. In Proceedings of the IEEE International Conference on Information and Automation (ICIA), Shenzhen, China, 6–8 June 2011; pp. 601–605.

Sensors 2016, 16, 931

15.

16.

17. 18.

19. 20.

21.

22.

23. 24. 25. 26. 27. 28. 29.

16 of 16

Salvadori, C.; Petracca, M.; Ghibaudi, M.; Pagano, P. On-board Image Processing in Wireless Multimedia Sensor Networks: A Parking Space Monitoring Solution for Intelligent Transportation Systems. In Intelligent Sensor Networks: Across Sensing, Signal Processing, and Machine Learning; CRC Press: Boca Ratón, FL, USA, 2012; pp. 245–266. Alessandrelli, D.; Azzarà, A.; Petracca, M.; Nastasi, C.; Pagano, P. ScanTraffic: Smart Camera Network for Traffic Information Collection. In Proceedings of the European Conference on Wireless Sensor Networks, Trento, Italy, 15–17 February 2012; pp. 196–121. Liu, J.; Chen, R.; Chen, Y.; Pei, L.; Chen, L. iParking: An Intelligent Indoor Location-Based Smartphone Parking Service. Sensors 2012, 12, 14612–14629. [CrossRef] [PubMed] Gu, J.; Zhang, Z.; Yu, F.; Liu, Q. Design and implementation of a street parking system using wireless sensor networks. In Proceedings of the 10th IEEE International Conference on Industrial Informatics (INDIN), Beijing, China, 25–27 July 2012; pp. 1212–1217. Reve, S.V.; Choudhri, S. Management of Car Parking System Using Wireless Sensor Network. Int. J. Emerg. Technol. Adv. Eng. 2012, 2, 262–268. Giuffrèa, T.; Siniscalchia, S.M.; Tesorierea, G. A novel architecture of parking management for Smart Cities. In Proceedings of the 5th International Congress—Sustainability of Road Infrastructures, Rome, Italy, 29–31 October 2012; pp. 1278–1287. Yang, J.; Portilla, J.; Riesgo, T. Smart Parking Service based on Wireless Sensor Networks. In Proceedings of the 38th Annual Conference on IEEE Industrial Electronics Society, Montreal, QC, Canada, 25–28 October 2012; pp. 6029–6034. Geng, Y.; Cassandras, C.G. A new “Smart Parking” System Infrastructure and Implementation. In Proceedings of the 15th meeting of the EURO Working Group on Transportation, Porto, Portugal, 4–6 September 2013. Tian, Q.; Guo, T.; Qiao, S.; Wei, Y.; Fei, W. Design of Intelligent Parking Management System Based on License Plate Recognition. J. Multimed. 2014, 9, 774–780. [CrossRef] Karunamoorthy, B.; SureshKumar, R.; JayaSudha, N. Design and Implementation of an Intelligent Parking Management System using Image Processing. Int. J. Adv. Res. Comput. Eng. Technol. 2015, 4, 85–90. Caballero-Gil, C.; Molina-Gil, J.; Caballero-Gil, P. Low-Cost Service to Predict and Manage Indoor Parking Spaces. In Lecture Notes in Computer Science; Springer: Heidelberg, Germany, 2015; Volume 9454, pp. 225–236. Texas Instruments Incorporated, CC2530 (Rev. B). Available online: http://www.ti.com/product/CC2530/ technicaldocuments (accessed on 27 March 2016). Geoff, M. The 6LoWPAN architecture. In Proceedings of the 4th Workshop on Embedded Networked Sensors, EmNets ’07, Cork, Ireland, 25–26 June 2007; pp. 78–82. SHARP Corporation, GP2Y0D02YK0F. Available online: http://www.sharp.co.jp/products/device/doc/ opto/gp2y0d02yk_e.pdf (accessed on 27 March 2016). Contiki-Developers, Contiki 2.6. Available online: http://contiki.sourceforge.net/docs/2.6/ (accessed on 27 March 2016). © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).