A Novel IoT-aware Smart Parking System based on the integration of RFID and WSN technologies

Abstract

Enabling a sustainable mobility is one of primary goals of the so-called Smart Cities vision, and in this perspective, the deployment of intelligent parking systems represents a key aspect. This paper presents a novel IoT-aware Smart Parking System based on the jointly use of different technologies, such as RFID, WSN, NFC, and mobile. It is able to collect, in real time, both environmental parameters and information about the occupancy state of parking spaces. To reduce the overall system costs, the possibility to use a solar RFID tag as cars’ detection system has been analyzed. The system allows drivers to reach the nearest vacant parking spot and to pay for the parking fee, by using a customized mobile application. Furthermore, a software application based on RESTful Java and Google Cloud Messaging technologies has been installed on a CS in order to manage alert events. A proof-of-concept has been defined to demonstrate that the proposed solution is able to satisfy real requirements of an innovative Smart Parking System, while preliminary analysis of solar tag usage investigates the feasibility of the proposed detection solution.

Keywords

RFID (Radio Frequency Identification)Wireless Sensor Network (WSN)Near FieldCommunication (NFC)mobile payment Internet of Things (IoT)smart cities Smart Parking System

1 Introduction

The recent progresses in Micro Electro-Mechanical Systems (MEMS), together with the widespread diffusion of Internet, has led to the assertion of the Internet of Things (IoT) concept, according to which, in a near future, the objects of everyday life will be able to communicate with one another and with the users, becoming an integral part of the Internet (Mainetti et al., 2011). By enabling easy access to a wide variety of devices such as home appliances, monitoring sensors and actuators, the IoT will foster the development of a plethora of applications, ranging in many different domains, such as home and industrial automation, healthcare and elderly assistance, intelligent energy management and smart grids, automotive, traffic management and many others (Catarinucci et al., 2015). In particular, the application of the IoT concept to the urban scenario is growing of interest as it responds to the strong requirements of many national governments to adopt Information and Communications Technologies (ICTs) in the management of public services, thus realizing the so-called Smart Cities concept (Zanella et al., 2014). In this perspective, the IoT can significantly contribute to the design of a smart mobility able to answer to users’ request in terms of transport network efficiency and social sustainability.

Reducing urban traffic congestion represents, in fact, a primary goal for most of the cities, and the development of intelligent management systems for public parking represents a fundamental aspect. It is estimated that 30% of the daily traffic congestion in an urban downtown area is caused by vehicles cruising for parking space, and that a driver spends on average 7.8 min to find a parking spot (Arnott et al., 2005). Furthermore, often drivers, frustrated by the lack of parking spaces, use the parking spots reserved for particular categories of people, such as disabled or law enforcement. This not only causes waste of time and fuel for drivers looking for parking but also increases air pollution and drivers’ frustration. Such considerations suggest the development of new smart parking systems able to better manage the urban parking areas and to perform road traffic issues.

Among all technologies enabling the development of innovative smart applications, Ultra High Frequency (UHF) Radio Frequency Identification (RFID), Wireless Sensor Network (WSN), and mobile represent three of the most promising candidates to face these challenges. RFID is a low-cost, low-power technology consisting of passive and/or Battery-Assisted Passive (BAP) devices, named tags, which are able to transmit data when powered by the electromagnetic field generated by an interrogator, named reader (Dobkin, 2012). Since passive RFID tags do not need a source of energy to operate, their lifetime can be measured in decades, thus making the RFID technology well suited in a variety of application scenarios (Mainetti et al., 2013). Nevertheless, the main drawback of RFID tags stems from the fact that they can operate solely under the reader coverage region, i.e. up to 10 m and 50 m when respectively fully-passive and BAP tags are used (Solic et al., 2012). Clearly, such an aspect limits the use of UHF RFID technology to object identification and monitoring within quite smallareas.

On the contrary, WSNs are basically self-organizing ad-hoc networks of small, cost-effective devices that communicate in a multi-hop fashion to provide monitor and control functionalities. Currently, most WSN motes are battery-powered computing platforms integrating analog/digital sensors and an IEEE 802.15.4 (IEEE802.15.4, 2011) radio enabling up to 100-m outdoor communication range (single hop). Let us observe that WSN motes consume significantly more power than RFID tags, thus making the overall network lifetime the major limitations of such technology. RFID and WSN represent two complementary technologies whose physical integration might provide augmented functionalities and give new perspectives to a broad range of innovative applications. Moreover, the adoption of widely recognized standards, such as 6LoWPAN, and Constrained Application Protocol (CoAP) (Bormann et al., 2012), an HTTP-like protocol especially designed for resource constrained devices, guarantees the development of extensible solutions, which can be easily integrated with a more complex Smart City infrastructure.

However, to the best of authors’ knowledge, only few attempts have been done to leverage the combined use of UHF RFID, WSN and mobile in urban mobility scenarios. Furthermore, none of the available solutions is based on the use of standardized IoT protocols. Therefore, adopting an innovative parking management system capable not only to drive users toward the vacant parking lots, but also to allow traffic authorities to adequately monitor the state of reserved parking spaces, could significantly improve the citizens’ quality of life. The ability for a user to automatically pay for the occupied parking space could be a very important feature, as it would allow the user to deal with a single application and, at the same time, it would enable traffic authorities to perform real time checks of parking fees.

In order to address these issues, in this work, a novel IoT-aware Smart Parking System (SPS) is presented and discussed. It is able to guarantee innovative services for the automatic monitoring and management of parking spaces, by exploiting the potentialities offered by the jointly use of different, yet complementary, technologies and standards, such as RFID, WSN, NFC, mobile, 6LoWPAN, and CoAP. In particular, the primary objective of this last one is to provide a lightweight access to physical resources in order to meet the limited capabilities of embedded devices. Specifically, CoAP allows sensor nodes to run embedded Web services, through which their resources can be easily manipulated. Furthermore, the development of Representational State Transfer (REST)- style IoT systems enables seamless interoperability with cloud services, thus guaranteeing the implementation of scalable solutions, which can be easily integrated with a Smart City infrastructure. The designed SPS is able to collect, in real time, both environmental parameters and information about the occupancy state of parking spaces via an ultra-low-power Hybrid Sensing Network (HSN) composed of 6LoWPAN nodes integrating UHF RFID Class-1 Generation-2 (Gen2) functionalities (EPCglobal, 2013). The system directs drivers to the nearest vacant parking spot by using a customized software application. In order to reduce the overall system costs and mitigate the battery replacement issue for WSN nodes, the proposed SPS uses a customized solar RFID tag to detect the car presence on a particular parking spot. The developed tag is based on the BAP RFID technology (Che et al., 2009). BAP tags use a small battery to power up the tag circuitry, thus making the communication with reader more reliable. However, once its battery exceeds, it performs like passive tag. More in detail, the proposed solution considers the parking slots equipped with solar tags. As the car is located on top of the tag, it blocks the required amount of solar energy to power it up, and the system recognizes the spot as occupied. In this way, information about the occupancy state of parking spaces can be easily retrieved by RFID readers scattered in the parking area and delivered to a control center, where an advanced monitoring application makes them easily accessible via a REST Web service. Moreover, the designed mobile application allows users to pay for the occupied parking space, leveraging an authors’ previous work called IDA-Pay (Mainetti et al., 2012). The revised version of IDA-Pay, included in the current work, takes fully advantage of the Mobile Proximity Payment (MPP) increasing trend, according to which more that 50% mobile phones in Italy will be NFC-enabled in 2017. Furthermore, in case of improper use of a reserved space or expiration of the purchased time, the designed system is able to promptly inform the traffic cops equipped with a smartphone connected to a small portable UHF RFID reader.

The paper is organized as follows. In Section 2, the related work is analyzed, while the architecture of the proposed SPS along with involved hardware and software components are outlined in Section 3. Section 4 discusses the adopted RFID-WSN integration strategy. Details on the implemented architecture are given in Section 5 while a prototype implementation of the proposed SHS is described and validated in Section 6. A qualitative comparison between our proposal and some smart parking solutions is reported in Section 7. Concluding remarks are drawn inSection 8.

2 Related works

Over the last years, several research works aiming at improving parking management have been proposed. Most of them are based on the use of intelligent Parking Guidance and Information (PGI) systems able to provide the drivers with information on the location and the availability of spaces in car parks and direct them to vacant parking lots. The proper operation of these systems is based on the use of sensors able to detect the presence of vehicles placed in the vicinity of parking spaces. The choice of the type of sensor to be used depends on several factors, such as the parking type, the installation location, the connection type (i.e., wired or wireless) and sensitivity to external factors. In general, the range of sensors used to this purpose is very wide, including acoustic (Cevher et al., 2009), infrared (Hussain et al., 1995), light and ultrasonic sensors (Idris et al., 2009). More recently, the use of magnetic sensors based on magneto-resistors have been proposed (Zhang et al., 2011; Jian et al., 2011). A different solution is presented in (Wu et al., 2009), where video cameras are used to collect information of vehicle parking fields. This solution, however, requires the deployment of a wired network, since camera sensors usually generate a large amount of data, difficult to transmit wirelessly.

The use of wired or wireless sensors to detect the occupancy state of parking spots is not without shortcomings. Wired sensors, indeed, are intrusive solutions, since they are typically installed in holes on the road surface by means of invasive procedures. On the contrary, wireless devices can be easily mounted on the ground or the ceiling of the car park. However, their use usually implies the development of complex mechanisms able to reduce the power consumption of the battery-powered devices. Moreover, the cost of commercial wireless sensors is still high; therefore, their use appears to be not suitable for large-scale deployments.

The adoption of vehicles detection systems based on the RFID technology represents the ideal choice to solve these problems, and to ensure the development of efficient and low-cost smart parking systems. The benefits of using such technology in smart parking applications were thoroughly analyzed in 2007 (Pala & Inanc, 2007). However, to date, RFID systems were primary used for the identification of vehicles entering into large parking areas (Wei et al., 2012), leaving the task to check the occupancy state of each parking space to different solutions. In our previous work (Mainetti et al, 2014), we proposed the use of UHF RFID tags to identify the unauthorized occupancy of reserved parking spaces, while adopting wireless sensor devices to detect the state of the parking spots. On the contrary, in order to reduce the overall system costs, along with the battery replacement issue for wireless sensor nodes, the adoption of solar RFID tags as detection sensors is suggested in this work. The idea of using the solar energy to power-up a tag, which is working as BAP tag, is not new (Sample et al., 2011), and some solutions are commercially available (GAO RFID, 2014). However, they use the solar energy to store the energy into super capacitor, and to use it for later operations. Our main idea is to utilize the enough light level to determine the car presence. Once the car is on top of the solar tag, the solar energy will not be enough to run the tag, thus denoting the slot as occupied.

In order to provide innovative services, an RFID-WSN integrated network can be used to retrieve and transmit real-time information about the occupancy state of parking spaces, together with data about ambient conditions (e.g., temperature, humidity and CO2 emissions). The deployment of WSN in parking lots has been extensively analyzed in the literature. In (Yang et al., 2012), authors used sensor nodes equipped with a light sensor to detect the state of each parking lot in an indoor area and to report the retrieved information to a Web server via the WSN. The information was also sent to a central server using a Wi-Fi network, and made accessible to the drivers through a mobile phone. In (Boda et al., 2007), authors proposed a detection scheme using magnetometer signature measurements able to track available parking spaces in public areas in real time and communicate the information to the users. Moreover, a new reservation-based smart parking system, which not only broadcast real-time parking information to the driver, but also provide reservation service was proposed in (Wang & He, 2011). Let us observe that, none of the available solutions realizes a seamless integration of different technologies, according to the IoT vision. The adoption of IoT-based communication standards, such as 6LoWPAN and CoAP, in urban scenarios represents an important aspect since it allows the development of innovative systems, which can be easily extended and integrated with a complete Smart City infrastructure. Furthermore, some CoAP built-in features, such as resource observation and discovery, enable a dynamic environment where the available resources are automatically discovered and configured.

Also, a number of commercial smart parking systems are available on the market. Smart Parking Systems is an Italian company, which proposes its FL200 (Smart Parking, 2014), a battery-assisted UHF RFID tag able to sense the car presence. Its battery life is around 8/10 years. An additional repeater module is used to extend the system’s coverage. The parking kiosk accepts the most commons payment methods. Park Smart (Park Smart, 2014) uses optical technologies to detect parked vehicles (reading their plate number via OCR). Siemens Integrated Smart Parking Solution (Siemens Urban Mobility, 2014) is a bigger platform, which is completed with an interesting forecast module (to suggest free parking spots probability basing on previous data). It uses the radar technology to detect the car presence. Kiunsys (Kiunsys, 2008) is a very complete Italian solution. It includes also a park booking feature and an RFID-based access control system for restricted areas. The Streetline solution (Streetline, 2014) uses 6LowPAN WSN for car sensing, which also allows to track sound level and surface temperature. It also provides public API to build custom apps. Fastprk (Fastprk, 2011) uses battery-assisted RFID tags. SmartCom (Smart Park, 2011) uses a mix of Infrared and HF RFID technologies to detect the car’s presence and to allow traffic agents verify parking permissions. Each of the proposed solution offers a mobile app both for drivers and traffic agents.

3 System architecture overview

This work aims at designing and implementing an IoT-aware Smart Parking System (SPS) having, as main peculiarity, the capability to seamless combine different innovative technologies. As general view, the designed system is able to collect, in real time, both environmental conditions and information about the occupancy state of parking spots and deliver them to a Control Center. At this point, an advanced monitoring application analyzes the received data and sends an alert message to the mobile application of the nearest traffic cop in case of unauthorized use of a reserved space or expiration of a parking receipt. Drivers can use a different customized mobile application to find the nearest parking spot and to pay the fee.

The conceived SPS has been put into effect according to the architecture illustrated in Fig. 1. To make the architecture description more readable, a list of acronyms used in the following has been reported in Table 1. As shown, it is composed of four main parts: (1) the RFID-enhanced wireless sensor network, named Hybrid Sensing Network (HSN) hereafter, (2) the IoT Smart Gateway, (3) the Central Server (CS), and (4) the user interfaces for data visualization and management.

The HSN consists of an integrated RFID-WSN 6LoWPAN network composed of two types of nodes: (i) 6LowPAN Border Router (6LBR), and (ii) 6LowPAN Router Reader (6LRR). According to the 6LoWPAN standard, the 6LBR is in charge of connecting the network to the Internet by translating 6LowPAN packets into IPv6 packets and vice-versa, while the 6LRR is defined as a 6LowPAN router node (i.e., a node able to provide forwarding and routing capabilities) interfaced with an RFID reader. More details about 6LRR capabilities are provided in the sub-section 4.1.

At a finer level of detail, the designed SPS assumes that customized UHF RFID Gen2 tags, equipped with a solar cell, are placed on each parking spot in order to detect cars’ presence, while 6LRR nodes are placed on poles located near the parking spaces. Once tags are provided with the enough light level, they transmits their ID back to 6LRR. The adoption of customized RFID tags as detection sensors, instead of WSN nodes, allows maintaining a high level of reliability, reducing, at the same time, the installation costs. The retrieved information is delivered, through the deployed WSN, to the IoT Smart Gateway. This last one is connected, on the one hand, directly with the HSN and, on the other hand, with the Internet through a 3G-communication interface. Therefore, in the proposed architecture, the gateway plays the role of 6LBR, enabling the communication between WSN nodes and remote users. The gateway, in turn, allows the RESTful communication with the CS.

The 6LRR nodes are also used to check that only authorized cars occupy the reserved parking places. These cars, in fact, are labeled with RFID tags containing information about their special permissions. In the near future, special RFID car plates could be used to store a number of information about the car and its owner, as well as optional special permission for impaired people or for granting access in traffic restricted areas. When the Management Application (MA) installed on CS realizes that a reserved parking spot has been occupied (i.e., the corresponding RFID tags was not read by 6LRR node responsible for controlling that specific reserved space), it checks if a new tag has been identified, and, in such a case, it verifies the car’s authorizations. For this purpose, the CS maintains a database storing a lot of information about parking spaces availability and user’s payments (Control DB in Fig. 1). To make the collected data easily accessible, the REST Web-based paradigm has been adopted. Specifically, a Web-based graphical interface has been developed in order to allow traffic cops and operators with specific privileges to access both real time and historical data about the state of parking lots.

Furthermore, the MA exploits Push Notifications (PN) to inform the traffic cop about the improperly use of a reserved space or expiration of parking receipt. Indeed, in the designed SPS the traffic cops are equipped with a smartphone connected to a portable RFID reader and running a customized application, named TrafficCop App. Through this App, traffic cops can interact directly with the tags placed on the cars and check their permissions by reading the information stored into the user memory of the RFID Gen2 tag or historical information stored into the Control DB. The TrafficCop App allows operators to issue a fine and to update the memory content of RFID tags with important information to remind (e.g. the date and time of the last check, information about the expiration of the authorization, etc.). A different mobile application, called ParkingApp, allows the driver to find the parking spaces available in a given area, get the right directions to the selected parking spot, pay the parking fee, check the remaining parking time and receive push-style notifications when the purchased time is expiring. Furthermore, the developed system allows the driver to pay for the parking service. Following the footsteps of one of our previous work called IDA-Pay, we avoid the user the pain of collecting coins for paying the due parking fee. The former IDA-Pay version used the NFC Peer-to-Peer (P2P) working mode to transfer payment credentials (e.g. encrypted credit card information, payment amount) to the Point of Sale (POS). The new version of IDA-Pay architecture works both with Android and iOS. The Android app uses the Host Card Emulation (HCE) technology that allows the mobile wallet to be compliant with the existing POS infrastructure. On the iOS side, pure card emulation is performed because of the presence of a local Secure Element (SE). We implemented a Cloud network for storing secure information and distribute virtual tokens when a payment request is forwarded.

Finally, in order to ensure an adequate level of security to data access and management, users (i.e., drivers and traffic operators) need to be authenticated before they can access the platform. Moreover, also local and remote communications must be adequately protected. In particular, in the latter case, it is necessary to provide a stronger communication channel, since the interaction between the remote application and the SPS is performed through the Internet. To do so, the proposed solution exploits a Virtual Private Network (VPN) channel that links the mobile device with the CS. Once this access is granted, the user can act on the system.

4 Hardware details

As discussed above, the envisioned reference scenario relies on the seamless use of IEEE 802.15.4-based WSN and RFID devices. Specifically, a 6LoWPAN router node and a RFID Gen2 reader have been physically interconnected in order to make up a 6LRR (Fig. 2). In such a way, the information exchanged between the tag and the reader via the Gen2 air interface become directly accessible by 6LowPAN devices, thus allowing standardized EPCglobal data to be relayed, in a multi-hop fashion, over the IEEE 802.15.4-based 6LowPAN network. Furthermore, the BAP RFID technologies has been adopted to develop the cars’ detection system, which represents a fundamental element of the proposed solution. Details about the hardware components making up the designed system are provided in the following.

4.1 Adding UHF RFID capabilities to 6LoWPAN nodes

As shown in Fig. 3 and accordingly to the architecture in Fig. 2, a 6LRR consists of a commercial RFID Gen2 Reader interfaced with the XM1000 mote from Advanticsys (Advanticsys, 2014) via the Universal Asynchronous Receiver/Transmitter (UART) communication bus. The XM1000 is based on “TelosB” technical specifications, with upgraded 116-Kb EEPROM, 8-Kb RAM and integrated temperature, humidity, and light sensors. It is equipped with a 16-bit ultra-low-power TI MSP430F2618 microcontroller unit (MCU). Wireless communication capabilities are provided by the IEEE 802.15.4-compliant TI CC2420 transceiver with transmission frequency of 2.4 GHz. The selected RFID Gen2 reader is the Sensor ID Discovery Gate UHF (Sensor ID, 2014), which can be easily configured and controlled by the XM1000 board via the UART interface. The RFID reader supports standard Read (Write) commands for reading (writing) data from (to) the RFID tag user memory via the Gen2 air interface. The reader operates in the standard European UHF RFID band (866–868 MHz) with a maximum equivalent isotropic radiated power (EIRP) of 2 W (33 dBm), adjustable via software.

The software aspects concerning the interfacing between XM1000 mote and Discovery Gate UHF RFID Reader have been implemented in Contiki OS (Dunkels et al., 2004), a popular open-source operating system targeted to small MCU-based architectures and developed by the Swedish Institute of Computer Science. Contiki OS communication stack is organized in several layers in which both protocol solutions and radio transceiver features can be thoroughly configured. Contiki OS provides a full IP network stack, with standard IP protocols such as UDP, TCP, and HTTP, in addition to new low-power standards. Specifically, Contiki OS supports also the recently standardized IETF protocols for low-power IPv6 networking, including the 6LowPAN adaptation layer, the RPL IPv6 multi-hop routing protocol, and the CoAP RESTful application-layer protocol. Furthermore, Contiki OS is highly memory efficient and provides a set of useful mechanisms for memory allocation. These features make Contiki OS the ideal choice for the development of new innovative smart applications, capable to exploit the new possibilities offered by the integration of RFID and WSN technologies.

Specifically, in the 6LRR node implementation, several functions allowing the XM1000 WSN mote to fully control hardware and software parameters of the Discovery Gate UHF RFID Reader, handle Gen2 inventory and Read/Write commands, manage the tag population, and retrieve data from Tags’ user memory via the UART communication interface have been implemented in Contiki OS as system driver.

4.2 BAP RFID solution

To detect the car presence in the parking slot, the usage of BAP RFID technology has been exploited. BAP tags use the battery to run tag circuitry, while the communication is the same as in the full passive RFID, where the tag ID is transmitted back to the reader by means of reflecting incoming radio waves and thus transmitting binary digits. Since car presence is only one bit of information, for this purposes we consider replacing the battery with the solar cell. In that way once the solar cell is lighten with enough amount of light, the tag is able to respond back to the reader with its ID number denoting the particular parking spot. On the contrary, once the light level drops below some level, the tag will stop transmitting and thus signalizing the car presence. The solution for tracing the parking slot occupancy is given in Fig. 4. For experimental purposes we used PowerID PowerG tag (PowerID, 2010), with battery replaced by Conrad yh-39×35 (SolarCell) solar cell. To experiment with reading capabilities we have used Alien ALR-9900 + RFID reader, with 6 dBi circularly polarized patch antennas (ALR-9900, 2012).

5 Software architecture details

The SPS is a complex system whose architecture is composed of several components. Among these, the block that guarantees the integration between RFID Gen2 and 6LoWPAN WSN technologies represents the core of the SPS In the following subsections, major implementation details on the software architecture of the SPS are provided.

5.1 Hybrid sensing network

In order to achieve a seamless interoperability between the HSN and Internet, the REST Request/Response paradigm piggybacked on CoAP messages has been exploited. CoAP is one of the most used communication protocol in the IoT and its primary objective is to provide a lightweight access to physical resources in order to meet the limited capabilities of embedded devices. CoAP design is similar to that of HTTP since it provides a request/response model interaction between two end-points and includes key concepts of the Web, such as URI and media types. In addition, CoAP provides a resource observation mechanism, which allows a client to receive notifications upon every change in the state of resources it has previously subscribed to.

As summarized in Table 2, two different kinds of resources can be identified in the proposed architecture: (i) environment sensors, and (ii) RFID-related resources. More in detail, 6LRR nodes placed on poles are equipped with several sensors able to monitor environmental parameters and, therefore, they expose CoAP ambient sensor resources (e.g. coap://[aaaa::1]/ambient/light and coap:// [aaaa::1]/ambient/temperature). However, such kind of nodes are also equipped with an RFID reader, which allows them also to expose an RFID resource (coap://[aaaa::1]/RFID/reader). This last one represents aggregated information of tags read within the 6LRR RFID range. In this way, each resource can be individually accessed from anywhere in the Internet by using CoAP methods.

For the sake of simplifying the development of a new class of services capable to exploit the new possibilities offered by the RFID-WSN integration, we drawn on the implementation of Erbium (Er) (Kovatsch et al., 2011), a low-power REST engine for Contiki. Such implementation has been adapted to our hardware. Specifically, in order to get sensor readings, each sensor has been registered as a resource and a proper handler for each sensor has been defined. Upon receipt of a GET request coming from client applications, the handler polls the sensor and builds the response message using the sensor state as payload. If the request is an observation request, the client is registered as an observer for future notifications. The registration of a single observer will trigger the activation of a function that periodically checks for resource state changes and informs all registered observers. In the proposed SHS, the observation of the RFID related resource on the 6LRR allows client applications to be automatically notified when the numbers of RFID tags read changes compared to a previous reading, showing that something new has happened in the parking lot (i.e, a new car parked or a car has gone on).

5.2 IoT Smart Gateway

The IoT Smart Gateway represents an important element of the designed SPS architecture. It is in charge of data collection and transmission, and payment execution. The IoT Smart Gateway has been realized by connecting a Rasperry PI board (Raspberry Pi Foundation, 2014), equipped with the Raspian OS, to the 6LowPAN Border Router. It has been also equipped with a GPRS module, a GPS module, and the SCL3711 multi-protocol 13.56 MHz contactless reader. Raspberry Pi is a credit card-sized computer powered by the Broadcom BCM2835 system-on-a-chip (SoC). This SoC includes a 32-bit ARM1176JZFS processor, clocked at 700 MHz, and a Videocore IV GPU. It is equipped with 512 MB of RAM and powered by a 5 V micro USB AC charger. From a functional point of view, it is mainly composed of two different blocks: a proxy and the billing subsystem.

5.2.1 Proxy

The Proxy enables transparent communication with CoAP devices. It has the burden of translating HTTP requests coming from user interfaces (i.e. web or mobile applications) and the MA into CoAP messages and vice versa. Specifically, the Proxy is able to receive, process, and reply to requests, in JSON format, coming from the MA and the user interfaces. It has been developed by using the Spring Framework and deployed on the Jetty application server installed on the IoT Smart Gateway. The proxy logic has been extended by implementing a caching service, thus supporting multiple requests to the same resource and limiting the amount of traffic injected into the IoT peripheral network. This feature is particularly important for constrained nodes, which are not able to simultaneously manage requests from multipleclients.

5.2.2 Billing system

An innovative billing system, based on advance payments that uses customer’s personal devices to forward billing information, has been designed. Mobile Proximity Payments (MPP) is the most promising payment service, with biggest growth rate for the next years. It is foreseen that in 2017 more than one out of two mobile phones will be MPP-enabled. MPP is safe, pleasant and convenient for costumers, hence its success.

NFC is the MPP-enabling technology. It allows a mobile phone to be used as it was a contactless credit card. When a NFC mobile phone falls within the short range (i.e., 2-3 cm) of an NFC Point-of-Sale (POS), an RF link is established between the two devices. Google is probably the top name in the NFC world, because of the effort pursued by the IT giant in providing strong, affordable and customizable NFC features in Android mobile phones since 2010. Also Apple in 2014 declared the iPhone 6 supports to NFC mobile payment through the Apple Pay platform. Despite the link-layer technology – NFC – is identical for both Android and iOS, the approach to security is completely different. Usually, cards are emulated by a separate chip in the device, called ‘Secure Element’ (SE). Apple has the full control on the hardware of its devices, Google does not. Hence, Android 4.4 (codename KitKat) introduces an additional method of card emulation that does not involve a SE, called Host-based Card Emulation (HCE). It allows any Android application to emulate a card and talk directly to the NFC POS. The emulated card can change at every payment, by the means of a Virtual Card Provider (VCP). Apple instead uses the SE to store tokens retrieved from a Token Service Provider(TSP).

The IDA-Pay NFC POS is embedded in the smart gateway, which also contains the logic to communicate with a third-party payment processor service. For this work, we have fatherly extended the IDA-Pay system in order to support both the novel HCE technology and Apple Pay. In Fig. 4, the billing subsystem architecture is depicted.

For users for Android phones, the ParkingApp triggers the IDA-Pay feature via the use of the Intent-Filter Android paradigm. For Apple users, an additional fingerprint recognition is required (through the built in iPhone TouchID feature). The IDA-Pay apps – both Android and iOS versions – can talk to the smart gateway exchanging Application Protocol Data Units. At the end of the payment process, a confirmation is packaged and the ParkingApp shows up back.

The billing subsystem is composed of four main entities: (i) the user’s NFC mobile phone, (ii) the smart gateway embedding the IDA-Pay POS terminal, (iii) two third-party payment processor services (VCP for HCE, and TSP for Apple Pay), and (iv) the proprietary payment networks.

The Raspberry board has a NFC reader attached via USB. The Raspberry runs the IDA-Pay POS Java application that can talk to the NFC reader using the PC/SC standard protocol. In Java, this protocol is implemented in the smartcardio package (available from Java 1.6). Using the PC/SC standard, the POS application sends a “SELECT AID” command to the NFC reader, passing an Application IDentifier as argument. In ISO7816-4, specifying an AID is mandatory to select the smartcard function to be used. Main Credit Card networks like Visa and Mastercard have their own AIDs. We used both of them in order our app to be compliant to the two most spread payment network. On Android, once the AID is selected, all the future messages exchanged between the POS and the mobile phone will be handled by the IDA-Pay app. In iOS, the Apple Pay iPhone feature is the bridge between the app and the POS. The Java application also embeds a SOAP client, which is able to interact with the Web services exposed by the third-party payment processor we selected for our prototype. Such interaction is needed in order to validate the credit card information retrieved via the NFC link and then to charge the user with the due fees. What happens between the Payment processor services and the proprietary payment networks (e.g. Visa) is a black box for our system.

5.3 Central server

The CS represents the core of the proposed architecture. It controls the overall SPS behavior. The different components of the CS are highlighted in Fig. 1 and described below:

Management Application and System DB. The MA is a Standalone Java Application, easily configurable and accessible via the user interfaces. It performs two different tasks: (i) allowing ambient operators to control environmental conditions, and (ii) monitoring the parking lots state and alerting traffic cops in case of critical situations. For these purposes, the MA stores the information retrieved from the HSN nodes on a MySQL database, called Control DB. The presence of this database decouples data collection from data processing and visualization, so operators do not need to directly interrogate HSN nodes during the normal mode of operation. More in detail, the MA registers itself as an observer to the RFID reader related resources exposed by 6LRR nodes scattered in the parking space. When the number of RFID tags read from 6LRR node changes (e.g., it increases/decreases due to the arrival/departure of a car), all the information stored into the memory of the tags inside the coverage region are delivered to the MA for a further analysis. Such conditional observation method allows the MA to be notified only when something new happens in the parking lot, thus substantially reducing the number of notification messages in the network. The MA is also able to send Push Notifications to the mobile devices of traffic cops. It is worth to note that the use of the Push Notifications instead of other technologies (e.g. GSM) allows the SPS to directly interface with the mobile App and, therefore, to provide all the information about the parking lots state and user’s payments stored into the database. In particular, we resorted to the Amazon SNS cloud service, since it can seamlessly scale and add an abstraction level allowing programmers to use the same APIs for sending notifications on different platforms (e.g. iOS and Android).

Secure Access Manager and User DB. The Secure Access Manager (SAM) application ensures privacy and data protection. It coordinates all communication between end users and CS, providing access to stored information only to authorized users, i.e. registered on the User DB.

5.4 User Interfaces

Authorized users can interact with the system through user interfaces, accessible via Web browser by both fixed workspaces and mobile devices. Specifically, such interfaces implement RESTful services, which allow user to communicate with the HSN through the Proxy. The developed user interfaces offer two main functionalities depending on two possible client profiles:

Traffic staff Interface. This interface allows traffic operators to manage current and historical information from environmental sensors, visualize and eventually change the data about parking lots state and users’ payments stored into the Control DB. Traffic cops can interact remotely with the system by using the TrafficCop App, a customized Android application. As previously described, the App also allows traffic cops equipped with an RFID-enabled smartphone to check the notifications related to the detection of an unauthorized use of a reserved space, directly reading the information stored in the RFID tag placed on a machine, and to issue any fines.

Driver Interface. This interface allows drivers to visualize information about the occupancy state of parking lots and the executed payments. Specifically, users can use a customizes Android application, called ParkingApp, to find the parking spaces available in a given area, get the right directions to the selected parking spot, pay the parking fee, check the remaining parking time and receive notifications when the purchased time is expiring. Let us observe that ParkingApp uses the payment services provided the by the IDA-Pay ecosystem. IDA-Pay uses a secure storage to retain and retrieve credit cards data. This data box is located on the usual smartphone memory, but it is secured using an encryption mechanism (as foreseen in IDA-Pay). Only the smart gateway contains the logic needed to understand the data retrieved from the App secure storage. The HCE dispatcher running in the Android OS knows that the IDA-Pay’s HCE service, listening for a precise AID, is up and running.

6 System validation

6.1 Detection system experimental results

In order to develop solar RFID solution that is able to detect car presence, one need to pick proper RFID tag, and adequate solar cell. The tag performance depends on its chip sensitivity, antenna performance and the impedance matching between tag and antenna, while solar cell performance depends on current-voltage characteristics under different amount of solar irradiance. Measuring tag performance is non-trivial, and is often described under anechoic chamber to reduce the number of radio effects such as multipath, shadowing and fading present in real communication channel (Catarinucci et al., 2011). Therefore, we rely to the commercial tag which impedance and antenna details are unknown to authors, and which battery is replaced by adequate solar cell (picked in the way to approximate required energy we can measure). Replacing the battery with adequate substitute for purposes of increasing the reliability of RFID system is already available in (Russo et al., 2013), and (Solic et al., 2013).

To analyze the feasibility of solar tag, some tests have been carried out by using two cells (2 V and 4 V(yh-39×35) cell) in particular scenarios, which are of the interest for car detection purposes. To provide a reference, and repeatability of the experiments a pyranometer, able to provide the intensity of sun radiation (W/m²) has been used. In order to analyze the cell performances, the measurements of both cells depicted in Fig. 5 have been performed in outside location, during the sunny day, and indoor environment (without the shades on window, and turned neon lights on). Measurement results are listed in Table 2.

The preliminary results on the possibility of car detection by using the solar tag are partially described in (Solic et al., 2015). To investigate the possibility of using the solar tag, the BAP tag (i.e., PowerID - Power P BAP tag) power characteristics have been firstly checked in order to determine under what circumstances the power provided by the cell is enough to run the tag. To this purpose, the tag has been connected to a laboratory power source and connected tag to the laboratory power source. In these experiments, the tag was placed in the indoor environment containing tables, chairs, wooden cabinets, and computer equipment. The reader antenna was mounted on the ceiling, while tag was located on the floor approximately 4 meters away from antenna, and under the angle of 60 degrees. To check the power requirements, the voltage setting has been slightly changed in order to evaluate the minimum value required to power the tag on and to be confidently read. In (Solic et al., 2015) it was shown that current-voltage and power requirements for tag to function properly are to be non-linear. Those are depicted in Fig. 6, the same picture represents solar cell outputs under different values of sun irradiance. The intersection between solar cell and tag characteristics are representing the system operating point, i.e. the voltage, and the current that tag is drawing from the solar cell. From Fig. 6 it can be seen that light level of below 10 W/m² is not delivering the enough power to run the tag, while the experiments showed that we need of about more than 10 W/m² to run the tag confidently, especially under the car position wherein the propagation radio waves is hard due to metal conditions. Values below this threshold turn tag off and thus signalize the parking slot occupancy. Therefore, we have considered a few measurements on the level of power which can be obtained using given solar cell at the particular scenario of sunny day, and total shadows. Moreover, we have provided a few measurements related to the indoor garage positions between car-light and pyranometer. This means that the detection system may be used in the garages, but the paradigm should be changed, i.e. solar tag should be located in the front of the car light, and once it works for some time, this means that the given parking slot is occupied. The results are given in Fig. 7, where it can be seen that under some circumstances during sunny day, and shadows, the system would fail in detecting the car presence if tag is located on the edge of the car (the situations above 10 W/m² plane). This may be an issue for smaller cars (the used one is of 3,97×1,63 m dimensions), where one should investigate the optimal solar tag position, or may use multiple solar tags to reduce the likelihood of false alarms. Further, the car represents the metallic obstacle, and its impact should be elaborated in the future work.

In both cases – low light indoor environments and night scenarios – introducing an application level status memory will help to retain the actual status even when it is not instantaneously predictable. Presented results are providing further insights on possible utilization of solar tag for purposes of parking slot occupancy detection. As this is only the preliminary analysis, the results are promising for further system evaluation.

6.2 Overall architecture functional validation

In this section, a prototype implementation of the proposed SPS is described and validated. A simple proof-of-concept has been defined in order to demonstrate the feasibility of the proposed system.

The considered scenario, depicted in Fig. 8, includes a driver with a special grant: a permission to stand the vehicle for an extra time after the parking time has expired (for example, for business needs). In this case, the traffic cop should recognize the driver’s grant and avoid issuing the fine.

In our implementation, the car is equipped with a passive UHF RFID tag containing the Electronic Production Code (EPC) and information about the grant. In the prototypal implementation the Alien ALN-9654 G RFID tags have been used. This choice has been mainly done due to their extreme low-cost and compliance with the EPC standard. Let us observe that the aim of this paper is to demonstrate the feasibility of just one of the several possible use-case scenarios of the proposed SPS. Therefore, finding the best tag to use to detect a car is outside the scope of this work. Furthermore, the Control DB stores information about the location of each 6LRR node in the parking area and data about the employed traffic cops while a Nexus 4 mobile phone running Android 4.4.3 “KitKat” connected to the BlueBerry RFID Gen2 reader from TERTIUM Technology is used as handled reader.

As shown in Fig. 8, each WSN node can expose a variety of resources which the IoT Smart Gateway connected to the 6LBR can manipulate through CoAP methods. Specifically, the main actor of the system is the 6LRR node, in charge of reading and delivering to the IoT Smart Gateway data retrieved from the memory of the RFID Gen2 tags. To this end, in the considered example, the 6LRR node exposes the following resources:

coap://[aaaa::1]/RFID/reader

coap://[aaaa::1]/ambient/light

coap://[aaaa::1]/ambient/humidity

coap://[aaaa::1]/ambient/temperature.

The considered use case relies on the following operations:

The driver uses the ParkingApp to find a not reserved vacant parking unit (Fig. 9a). The App drives the user to the selected parking space (Fig. 9b). The 6LRR detects the car’s presence (i.e., the number of RFID tags read is changed).

The 6LRR sends a notification message containing the read data to the IoT Smart Gateway using a CoAP method.

The IoT Smart Gateway transmits the retrieved information to the CS, where the MA analyzes the received data and stores them into the Control DB.

The driver selects the amount of parking time (e.g., 1 hour) and click on ‘Pay’ button in the ParkingApp (Fig. 9c). The app asks the user to scan the bar code stuck onto the smart gateway. This sends the right fee to the right smart gateway, so the embedded POS can initiate the transaction. The driver touches his/her device to the smart gateway in order to settle the transaction. For Android phones, the IDA-Pay app shows app. For Apple phones, the PassBook app shows up.

After 1 hour the bought time expires, but the RFID reader embedded in the 6LRR does not still read the corresponding RFID tag. Hence the traffic cop mobile phone gets notified about the parking unit to check (Fig. 10a).

The traffic cop reaches the parking unit (Fig. 10b) and identifies the driver by reading the car’s RFID tag (Fig. 10c). The traffic cop confirms to the CS that the user is allowed to stand his/her car for an extra time because of a special permission.

7 Discussion

As highlighted by the state-of-the-art analysis proposed in Section 2, only few attempts to combine emerging IoT enabling technologies in the urban mobility application scenario have been proposed in the literature. Therefore, we decided to compare the proposed SPS with existing commercial solution rather than with academic works because of the completeness of our solution in terms of number of features available.

With respect to such works, the main advantage of our solution consists in its intrinsic scalability to large-scale deployment. This important feature is guaranteed by the use of IoT standards, such as 6LoWPAN and CoAP. The benefits provided by the adoption of such standards have been recognized by the research community and the general consensus is that they represent the foundation for the development of complete Smart Cities infrastructures (Zanella et al., 2014). It is worth to notice that the use of standard protocols, instead of proprietary and closed solutions, makes the designed architecture easily interfaceable with different technologies and third-party systems.

Furthermore, as clarified in Sections 3 and 5, the proposed SPS is able not only to monitor parking spaces but also to provide drivers and traffic cops with advanced features and services.

A comparison overview between our SPS and some of the most relevant systems in the market is presented in Table 4. Let us observe that all the considered works provide basic features, such as parking spots monitoring and parking guidance services.

Another very common service is the parking fee payment. However, the proposed solution is the only one exploiting NFC HCE to create the driver’s digital wallet. In this perspective, although most of the solutions on the market provide the authority with the capability to identify non-paying cars, only few works can properly manage the improper use of reserved parking spaces. This last feature requires, in fact, the car’s identification though the use of RFID technologies.

Finally, differently by the commercial solutions, the designed SPS is able to collect in real time, not only parking related information, but also environmental parameters and to direct them to a control center. Such data could be easily accessed and used by administrative authority to provide citizens with several interesting services.

8 Conclusions

In this paper, a novel IoT-aware Smart Parking System, able to reduce the traffic congestion and improve the citizens’ quality of life, has been presented. By exploiting the jointly use of different, yet complementary, technologies and standards, the proposed system is able to collect, in real time, both environmental parameters and information about the occupancy state of parking spaces. To this purpose, a Hybrid Sensing Network has been deployed. It is able to combine the communication capability of a WSN, with the identification features of the RFID technology. Furthermore, to provide advanced services, two different Android mobile apps, called ParkingApp and TrafficCop App, have been designed. By using the ParkingApp, a driver can find the parking spaces available in a given area, get the right directions to the selected parking spot, pay the parking fee, check the remaining parking time, and receive notifications when the purchased time is expiring. On the contrary, traffic cops, equipped with a smartphone connected to a portable UHF RFID reader, uses the customized TrafficCop App to interact directly with the RFID tags placed on the cars and check their permissions, or retrieve information stored into the cloud. The ParkingApp includes an NFC-based e-wallet system, which exploits the Host Card Emulation technology to allow users to pay for the parking fee, while leaving the user’s credit card information in the VISA and MasterCard cloud. Furthermore, we resorted to the Amazon Simple Notification Service in order to send alert to drivers and traffic cops. To reduce the overall system costs, we have considered to employ compatible RFID Gen2 RFID tags equipped with the solar cell to detect the car presence. Once the car is located on the top of solar tag, it will stop working due to insufficient light and thus signalizing the parking slot as occupied. During night scenarios we considered to change the paradigm, i.e. once the solar tag is enlighten by a car light, the solar tag is working and thus signalizing the occupancy. The preliminary results, presented in this paper are promising, and the future work will include the complete evaluation of such system, performed by considering a real development.

References

Advanticsys. (2014). AS-XM1000 resource manual. Retrieved from http://www.advanticsys.com/.

ALR-9900 (2012). Alien RFID reader, ALR-9900+RFID datasheet. Retrieved from http://www.alientechnology.com/.

Android. (2003). Android Operating System, Retrieved from http://www.android.com.

Arnott

, Rave

, & Schob

(Eds.). (2005). Alleviating Urban Traffic Congestion, Cambridge, USA: MIT Press.

Boda

V.K.

, Nasipuri

, & Howit

(2007), Design considerations for a wireless sensor network for locating parking spaces, Proceeding of IEEE SoutheastCon, 698–703.

Bormann

, Castellani

A.P.

, & Shelby

(2012). CoAP: An Application Protocol for Billions of Tiny Internet Nodes, IEEE Internet Computing, 16(2), 62–67.

Catarinucci

, De Donno

, Guadalupi

, Ricciato

, & Tarricone

(2011). Performance analysis of passive uhf rfid tags with gnu-radio, IEEE International Symposium on Antennas and Propagation (APSURSI), 541–544.

Catarinucci

, De Donno

, Mainetti

, Palano

, Patrono

, Stefanizzi

M.L.

, & Tarricone

(2015). An IoT-Aware Architecture for Smart Healthcare Systems, IEEE Internet of Things Journal, doi: 10.1109/JIOT.2015.2417684

Cevher

, Chellappa

, & McClellan

(2009). Vehicle speed estimation using acoustic wave patterns, IEEE Transactions on Signal Processing, 57(1), 30–47.

10.

Che

, Yang

, Xu

, Yan

, Tan

, Li

, Min

, & Tan

(2009). Analysis, Design and Implementation of Semi-Passive Gen2 Tag, Proceeding of IEEE International Conference on RFID, 15–19.

11.

Dobkin

(2012). The RF in RFID, Elsevier, 2012.

12.

Dunkels

, Gronvall

, & Voigt

(2004). Contiki – a lightweight and flexible operating system for tiny networked sensors, 29th Annual IEEE International Conference on Local Computer Networks, 455–462.

13.

EPCglobal, (2013). EPC Radio-Frequency Identity Protocols Generation-2 UHF RFID Specification for RFID Air Interface Protocol for Communications at 860 MHz-960MHz Version 2.0.0. Retrieved from http://www.gs1.org.

14.

Fastprk, (2011). The easiest way to park. Retrieved from http://www.fastprk.com/.

15.

GAO RFID. (2014). Solar powered UHF Gen2 RFID tag. Retrieved from: http://www.gaorfidassettracking.com/RFID_Asset_Tracking_Products/PDF/116421.pdf

16.

Hussain

, Baig

, Saadawi

, & Ahmed

(1995). Infrared pyroelectric sensor for detection of vehicular traffic using digital signal processing techniques, IEEE Transactions on Vehicular Technology, 44(3), 683–689.

17.

Idris

, Tamil

, Noor

, Razak

, & Fong

(2009). Parking guidance system utilizing wireless sensor network and ultrasonic sensor, Information Technology Journal, 8(2), 138–146.

18.

IEEE 802.15.4. (2011). IEEE Standard for Local and metropolitan area networks - Part 15.4: Low-Rate Wireless Personal Area Networks (LR-WPANs). doi: 10.1109/IEEESTD.2011.6012487

19.

Jian

, Hongbing

, Jie

, & Haitao

(2011). Data fusion for magnetic sensor based on fuzzy logic theory, Proceeding of International Conference on Intelligent Computation Technology and Automation (ICICTA), 87–92.

20.

Kiunsys, (2008). Kiunsys Innovation in mobility. Retrieved from http://www.infomobilitysuite.com/en/home.

21.

Kovatsch

, Duquennoy

, & Dunkels

(2011). A Low-Power CoAP for Contiki, IEEE 8th International Conference on Mobile Adhoc and Sensor Systems (MASS), 855–860.

22.

Mainetti

, Mele

, Patrono

, Simone

, Stefanizzi

M.L.

, & Vergallo

(2013). The impact of RF technologies and EPC standard on the fresh vegetables supply chain, International Journal of RF Technologies: Research and Applications, 5(1-2), 1–40.

23.

Mainetti

, Palano

, Patrono

, Stefanizzi

M.L.

, & Vergallo

(2014). Integration of rfid and wsn technologies in a smart parking system, Proceeding of 22nd International Conference on Software, Telecommunications and Computer Networks (SoftCOM), 104–110.

24.

Mainetti

, Patrono

, & Vergallo

(2012). IDA-Pay: A secure and efficient micro-payment system based on Peer-to-Peer NFC technology for Android mobile devices, Journal of Communication Software and Systems, 8(4), 117–125.

25.

Mainetti

, Patrono

, & Vilei

(2011). Evolution of wireless sensor networks towards the Internet of Things: A survey. In IEEE (Ed.) Proceedings of 19th International Conference on Software, Telecommunications and Computer Networks, 16–21.

26.

Pala

, & Inanc

(2007). Smart parking applications using rfid technology, Proceeding of 1st Annual RFID Eurasia, pp. 1–3, DOI: 10.1109/RFIDEURASIA.2007.4368108

27.

Park Smart (2014). Park Smart. Retrieved from http://www.parksmart.it/.

28.

PowerID (2010). Bap tag power-g datasheet. Retrieved from http://www.power-id.com/Products/PowerG.aspx.

29.

Raspberry Pi Foundation. (2014). Raspberry Pi resource manual. Retrieved from http://www.raspberrypi.org.

30.

Russo

, Šolić

, & Stella

(2013). Probabilistic modeling of harvested GSM energy and its application in extending UHF RFID tags reading range, Journal of Electromagnetic Waves and Applications, 27(4), 473–484.

31.

Sample

A.P.

, Braun

, Parks

, & Smith

J.R.

(2011). Photovoltaic enhanced UHF RFID tag antennas for dual purpose energy harvesting, Proceeding of 2011 IEEE International Conference on RFID (RFID), 146, 153.

32.

Sensor ID. (2014). Discovery Gate UHF Datasheet. Retrieved from http://www.sensorid.it/.

33.

Siemens Urban Mobility, (2014). Integrated Smart Parking Solutions. Retrieved from http://www.siemens.com/mobility/smart-parking.

34.

Smart Park. (2011). Smart Parking. Retrieved from http://www.smartparking.com/.

35.

Smart Parking, (2014). Smart Parking System. Retrieved from http://www.smartparking.it.

36.

SolarCell, Conrad yh-39x35 datasheet. Retrieved from: http://www.sciencetronics.com.

37.

Solic

, Radic

, & Rozic

(2012). Software defined radio based implementation of RFID tag in next generation mobiles, IEEE Transaction on Consumer Electronics, 58(3), 1051–1055.

38.

Solic

, Bozic-Kudric

, Russo

, & Rozic

(2013). Extending the reading region of the RFID UHF Gen2 system by HF/UHF integration, Telecommunications (ConTEL), 2013 12th International Conference on., 257–262.

39.

Solic

, Marasovic

, Stefanizzi

M.L.

, Patrono

, & Mainetti

(2015). RFID-based efficient method for parking slot car detection, 23rd International Conference on Software, Telecommunications and Computer Networks (SoftCOM), 108–112.

40.

Streetline. (2014). Streetline connecting the real world. Retrieved from http://www.streetline.com/.

41.

Wang

, & He

(2011). A Reservation-based Smart Parking System, IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS), 690–695.

42.

Wei

, Wu

, Yang

, Ding

, Li

, & Gao

(2012). Design and implementation of smart parking management system based on rfid and internet, Proceeding of International Conference on Control Engineering and Communication Technology (ICCECT), 17–20.

43.

, Huang

, Wang

, Chiu

W-C

, & Chen

(2007). Robust parking space detection considering inter-space correlation, Proceeding of IEEE International Conference on in Multimedia and Expo, 659–662.

44.

Yang

, Portilla

, & Riesgo

(2012). Smart parking service based on Wireless Sensor Networks, Proceeding of 38th Annual Conference on IEEE Industrial Electronics Society (IECON), 6029–6034.

45.

Zanella

, Bui

, Castellani

, Vangelista

, & Zorzi

(2014). Internet of Things for Smart Cities, IEEE Internet of Things Journal, 1(1), 22–32.

46.

Zhang

, Wang

, & Cui

(2011). Real-time traffic monitoring with magnetic sensor networks, Journal of Information Science and Engineering, 27(4), 1473–1486.