DOI:10.32604/cmc.2021.015265
| Computers, Materials & Continua DOI:10.32604/cmc.2021.015265 | |
| Article |
A Novel Green IoT-Based Pay-As-You-Go Smart Parking System
1Faculty of Information & Communication Technology, University of Malta, Msida, Malta
2Birla Institute of Technology, Mesra, Jharkhand, 814142, India
*Corresponding Author: Lalit Garg. Email: lalit.garg@um.edu.mt
Received: 09 November 2020; Accepted: 12 December 2020
Abstract: The better management of resources and the potential improvement in traffic congestion via reducing the orbiting time for parking spaces is crucial in a smart city, particularly those with an uneven correlation between the increase in vehicles and infrastructure. This paper proposes and analyses a novel green IoT-based Pay-As-You-Go (PAYG) smart parking system by utilizing unused garage parking spaces. The article also presents an intelligent system that offers the most favorable prices to its users by matching private garages’ pricing portfolio with a garage’s current demand. Malta, the world’s fourth-most densely populated country, is considered as a case of a smart city for the implementation of the proposed approach. The results obtained confirm that apart from having a high potential system in such countries, the pricing generated correctly forecasts the demand for a particular garage at a specific time of the day and year. The proposed PAYG smart parking system can effectively contribute to the macro solution to traffic congestion by encouraging potential users to use the system’s services and reducing the orbiting time for parking.
Keywords: Smart parking; garages; smart city; internet of things; traffic congestion; green IT; IoT
These days most of the cities in the world are struggling to solve a significant issue of vehicular parking, thanks to the scarce parking spaces and the rapid growth in the use of a personal vehicle due to the population growth, urbanization, and the bottleneck in public transport systems [1–4]. A green IoT-based smart parking system can ensure the best utilization of limited parking spaces. It will help in tackling traffic congestion [5], improving road safety [6–8], reducing pollution, minimizing the negative impact on city landscape and environment [9–11], supporting economic activities, tourism, and facilitating access and movement by pedestrians, cyclists, and mobility-impaired people [12]. Reference [10] discusses how individuals cause traffic congestion by traffic queuing caused by cruising for a parking place. Hence, our study considers how utilizing unused garage spaces for car parking can reduce the orbiting time for finding parking spaces. In this paper, we propose a novel smart parking system based on green IoT devices to manage, in the most beneficial way, the resource of unused garage spaces by taking Malta as a case paper. This uneven ratio between the lower increase in road infrastructure and the higher traffic increase will doubtlessly cause traffic congestion. We design an intelligent demand-forecasting pricing (IDFP) algorithm that uses geospatial data to achieve each garage’s most favorable price. To reach this objective, we researched and analyzed existing parking applications and solutions, which have already been successfully implemented in the local and international markets. Also, we evaluated current pricing methodologies, which were also put into effect in their scenarios. Although this research provided a background for a better design of the system by considering the desired pricing strategy, we had to research concrete market-based criteria on which such an economic-centric system would be based locally. Using a survey, we collected data about the proposed system’s potential market, which evolved around the Maltese scenario, and determined the cost of traveling to be used in a cost-benefit analysis of the end-product in the concluding stages. The survey also contained questions about a reasonable equilibrium market price for one hour of using such a system. The responses also showed in a significant way how various factors can fluctuate the demand and supply of the system and provided the necessary empirical economic data for the system’s design and the underlying intelligent pricing algorithm (IPA). Finally, we developed the API-centric Pay-As-You-Go (PAYG) smart parking system that constitutes the web application programming interface and the web and mobile applications. Along with the demand-forecasting, location-based IPA governing the system’s pricing strategy, we also designed and implemented the scheduling and fee quotation generation algorithms necessary to provide a complete PAYG smart parking system.
The next section reviews various existing parking systems and pricing strategies. This is followed by the data collection and analysis section. The System Design and Implementation section describes the proposed approach’s details, together with various algorithms implemented. Then, the Testing, Evaluation, and Results section present the results of manual testing of the proposed system’s algorithms and usability testing, followed by the performance and cost-benefit analysis. Finally, the conclusion section summarizes the contribution and offers ideas to extend the work in the future.
Parking spaces are rivals as they can only be used once by a single person at a specific time. Therefore, marginal cost pricing should be applied to parking. However, in Europe, many car drivers park for free or for a price far below the marginal cost they are incurring, resulting in increased parking pressure on these areas [1,12]. In several European cities, differentiated parking fees being implemented whereby parking fees vary according to the location, time, and/or vehicle type. The main aim is to make efficient use of the limited parking capacity. Other additional fees may be seen in places all around Europe. The electric vehicle can park for free, hybrid cars having a 20% reduction while heavy polluting vehicles pay an extra 20% [12]. Fees are the highest in the Central Business District of a city and decrease gradually [12].
2.2 Existing Parking Applications
During the last few years, several case studies worldwide emphasized the need for better traffic management with a particular reference to improve parking situations. [13] proposed a smart parking system where a parking service provider is selected based on service quality. Reference [14] proposed an IoT-based based system to facilitate interoperability between various parking service providers. References [15–19] review IoT devices and techniques for smart parking systems. Ensuring users’ security and privacy is essential, and therefore the use of blockchain technology is proposed in [20–22]. To find an empty parking space in our proposed green IoT smart parking system, we use Low-Power Wide-Area (LPWA) sensors to harvest the surrounding energy to get power for wake-up circuitry, as suggested by [17]. Therefore the battery would work for many years. Furthermore, we used Simtech’s Long Range (LoRA) protocol for the LPWA sensors, which is the most power-efficient [17].
There are a large number of parking systems, including SpotHero (Est: 2011) (US) [23] targeting commuters willing to park their vehicles in various garages, mainly multi-story car parks-enrolled in their system; PayByPhone (Est: 2001) [24,25], and on-street parking-based application; Parkopedia (Est: 2007) (UK) [26] focusing on PAYG on-street parking, and iParq [27], an intelligent community parking system. Other systems discussed literature include SFPark [9,28], ParkMobile [29], UParking [30], Parker [31], BestParking [32], ParkingPanda [33], mawgif [34], Mawqifi [35], ParkMe [36,37], and JustPark [38,39]. Tab. 1 analyzed some of these parking systems for developing our proposed smart parking systems’ pricing strategy.
Table 1: Summary of parking applications
2.3 The Economics Behind Road Space and Garage Parking
Reference [40] analyzed the factors affecting the market, i.e., the demand and supply of garage parking. This section will ultimately, together with the pricing strategies mentioned before, determine the economic concept which will govern the pricing strategy, thus the IPA of the developed system in this paper.
2.3.1 The Demand for Road Space
The demand for road space is derived demand as it is highly dependent on the demand for transport. Therefore, the higher the citizens’ willingness to pay to use their transportation, the higher the need for road space to arrive at their respective destinations.
In the above context, the demand for road space is highly dependent on the user’s destination. Due to this, [41] proposed pricing strategies and used in Rethymno, Crete, [42] to consider the user’s destination in the driver’s cost function and the agreed price with commuters, respectively.
A subtle determinant to the citizens’ demand for road space is their respective income, i.e., Car ownership positively correlates with the country’s yearly per capita GDP (Gross Domestic Product). This results in an unfavorable ratio between road infrastructure and traffic.
With one’s income, one of the two determinants of the driver’s cost function is the actual cost of transport, primarily fuel prices. An increase in the price of fuel triggers an inverse effect on the demand for transportation. However, the price elasticity of demand for road fuel in the short run is inelastically ranging between −0.15 and −0.28 [40]. This implies an increase in fuel or transport costs will trigger a minor decrease in a reasonable complementary price.
2.3.5 The Demand for Garage Parking
In addition to the above factors, the demand for garage parking is seen as also specifically having the following determinants:
i) The parking fees charged for using the garage parking service, also economically known as the price of the commodity itself
ii) The driver’s cost function of proximity to one’s destination against the fee charged
iii) The seasonality, i.e., the demand for a particular garage, varies according to either a meteorological or festive season affecting the garage’s geographic area.
2.3.6 The Supply for Garage Parking
Finally, we will now analyze the factors determining the supply of garage parking spaces. Garage Price-First of all, the price garage owners will be charging to the consumer is the leading contributor to this supply as this will be the owner’s primary revenue. Any extra commissions the owners need to pay to any intermediary company for their garages’ advertisement may negatively affect their willingness to put their garages on the market. Garage Location—In line with garage parking demand, location is also a leading determinant for garage parking supply. An owner will be more willing to make his garage available if he/she knows that the garage is in a strategic location near the main road or near commercial activity, especially at a particular time of year. Garage Features—a garage can earn higher revenue due to higher capacity, better maneuverability, and extra features/services as enhanced security or a valet. Reductions—Any subsidies or tax concessions given to the owners for offering such a service to help reduce the traffic congestion problem in a particular country will encourage them to make their resources available.
2.4 Existing Pricing Methodologies
In Tab. 2, we discuss four different cases of how parking systems utilize pricing strategies. The first of these cases is local, Valletta’s Controlled Vehicular Access (CVA) [43]. The other three are international: First, the dynamic demand-based pricing as applied in San Francisco in the innovative SFPark system [9,28]; second, the Contingent Valuation Methodology as used in Rethymno, Crete [42] and the third pricing system [41] is based on the driver’s cost function. This comparative analysis between different pricing strategies will help decide the most viable pricing strategy for the system proposed by this paper.
Table 2: Summary of pricing strategies
3 Data Collection and Analysis
To analyze the local market, i.e., Malta, we collected the realistic market data primarily to calculate an actual market base rate and its underlying factors and the importance of the system’s demand and supply sides separately to the main price determinants for garage parking. These were then be used by the dynamic demand-forecasting algorithm. Moreover, this survey’s results guided the overall design of the application prototype and helped to achieve a system tackling various areas accordingly based on Malta’s scenario. This survey’s targeted population was anyone over 18 years old, owning a driving license to match those who can potentially be stakeholders in this system. There were 194 participants with an equal distribution of age and income—and a dispersed locality distribution.
It may be seen from our survey that the private car is by far the way to go for more than 80% of the driving-eligible population, increasing to 85% on the weekend days. This contrasts with 38% and 68% using their private car on weekdays and weekend days, respectively, in a 2008 survey [44]. This increase in car usage matches well with the previously-mentioned correlation between GDP and car ownership.
As expected, an absolute majority of 72.8% of this survey’s respondents spend more than seven hours in productive work on weekdays. There is a remarkable similarity between the amount of time spent traveling and that for free-time activities on weekdays, with 22.1% and 24.1% of this survey’s respondents spending more than an hour in these activities. Moreover, on weekdays, 54.7% and 52.9% spend more than 1 or 2 h in free-time activities and traveling, respectively. Those traveling more than an hour on weekends may find a negligible increase in traveling time. However, almost 70% of the sample population spend between 45 and 2 h traveling on weekends. Tackling such an opportunity cost between traveling time and free-time activity time maximizes individuals’ time utilization.
3.3 Time Cruising for Parking or Stuck in Traffic
As per the above results, at weekends, although vehicles on the road increase, there is a notable decrease in the time spent stuck in traffic. Whereas 27.5% of the respondents spend more than 30 min a day stuck in traffic during the week, 37.2% of respondents spend more than 15 min during the weekend instead. Such a decrease is expected as although there are more cars on the road on weekends; however, they are less likely to go to the same place at the same time, mainly due to the absence of weekly school and work travel. In a PAYG smart parking system, pricing should vary in weekend day peak hours and those on weekdays where the highest activity is seen. Similarly, 74.4% of the respondents spend between 5 to 15 min. Specifically, 36.5% of respondents spend more than 15 min cruising for parking on weekends than 41.1% who, instead, spend more than 5-min cruising for parking during the week. If considering the thin margin between traveling time and other activities, especially during the week, such a 5-min daily delay may indeed make a difference. It will also be appropriate to minimize the more than 15-min delay found during weekends to utilize time better and reduce the monetary cost associated with fuel consumption.
3.4 Analyzing the Driver’s Cost Function
Further enhancing the theory mentioned in Geng et al. [41], our survey showed a 63.7:36.3 ratio in favor of a user’s proximity to the destination rather than the resulting cost. In the mentioned work, the researchers proved that pricing should be charged according to an individual’s cost function combining the location’s proximity to the individual’s destination with the cost associated with that. Although parking prices should reflect the proximity to a user’s destination as commercial areas, the cost associated should also be considered due to the notable 36.3% of respondents who are sensitive to the cost. As the latter cost includes the fuel consumption related to the cruising for parking, parking systems may charge a marginal increase to compensate for the time and monetary cost their service will save.
3.5 A Pay-As-You-Go (PAYG) Smart Parking System in Malta—The Demand
Primarily, 56.9% of the respondents responded that they would use such a system for a maximum of thrice a week. Such a figure is acceptable as these may either; prefer to go to work by public transport, choose to cruise for parking rather than paying every day, or already have a parking space provided for them by their employer. Thus these are likely to use such a system at weekends on their weekend travel. Special offers for more extended stays may be offered to attract such a population. On the other hand, 19.1% of the respondents are potential every weekday users. A combined 47.3% of the respondents will have their stay lasting longer than 2–3 h. On the other hand, 19% will have their stay lasting between 6–8 h, which matches those every weekday users mentioned above. Also, responses included those demanding the possibility of night stays and a more extended stay for three straight days.
3.5.1 A Pay-As-You-Go (PAYG) Smart Parking System in Malta—The Supply
Thirty-six respondents, or 19.6% of the sample size, responded as owners of garages falling under any mentioned categories. In the Household Budgetary Survey [45], it was found that, in 2008, 43000 dwellings were potentially unused. From our survey, with a confidence interval of 
5.59% from 19.6%, it was found that a minimum of 36600 and an average of 51150 garages are nowadays potentially unused or may have their space better utilized at various times of the day or week. The following results are based on a representative fraction of the respondents. We thus understand that further analysis should be made to arrive at a more precise conclusion from the whole 51150 potential suppliers. Nevertheless, these results will still give our article enough representative data to base this paper’s findings. Each garage’s capacity varies between 52.8% having one car, 33.3% having space for two vehicles, and another 8.3% having three parking spaces. This implies that if there are 36,600 garages, then 46,739 car parking spaces will be available, leaving 11339 car parking spaces available for those not using garage parking while at work. 45.9% of the respondents consider their garage as being in a ‘hot zone,’ and thus, they will be more willing to put their garage on the market via a PAYG smart parking system. 22.6% of the respondents said they would put their garage on the market depending on how much revenue they will generate, while 71.9% of the respondents said they would be willing to either free up space or refurbish the garage to make way for such a system. Whereas 18.8% of the respondents agree with a policy whereby they will be paid according to the sales generated, 81.3% of the respondents prefer a method whereby they will rent the garage and be paid a fixed amount for a defined period. Such figures should be considered when carrying out a thorough business analysis and design of a PAYG smart parking system. It should also be noted that the absolute majority of these respondents will only be willing to pay any commissions to a central company or pay for any security installations depending primarily on the monetary amount.
A third of the respondents were aware of at least one unused garage in their vicinity. Due to a secure parking spot with a minor charge, more than half of the respondents believe that such a system may contribute towards car-pooling. As such system will include a mobile application to allow maximum availability of its services; an Android app is preferred, with 63.5% of the respondents owning an Android smartphone, whereas another 22.8% of respondents owning an iOS smartphone. Nevertheless, an iOS counterpart or a hybrid application would provide the latter kind of users with maximum availability of the system’s services.
4 Calculating the Market’s Base Rate
As shown in Tab. 3, a weighted average hourly fee from both the supply and demand sides is derived from both high-demand and low-demand regions to estimate the market’s base rate. It was also based on the owners’ perception of their garages as in a ‘hot zone’ or not. The system’s demand-forecasting algorithm will use the resulting rate of €1.69 per hour as the base rate, which will then vary according to an individual garage’s demand. The 29% upper-bound would be used later on when assigning the actual weighting to different market factors to avoid over-inflating the resulting suggested rate per hour.
Table 3: Calculating the market’s base rate
4.1 Calculating the Market Factor Weights
In the survey, respondents ranked the significance of six market factors on their willingness to pay for a particular parking spot in a specific garage. Each rank was quantified with a value to arrive at a demand-side weighted average for each market factor:
a) The supply-side market factor weighting will use a similar strategy.
b) Finally, the whole market will weigh each market factor as the averaged weighted mean of the demand and the supply (As shown in Tab. 4).
Table 4: Calculating the market factor weights
*These are the highest possible weights to give to each market factor in the intelligent pricing algorithm.
5 The System Design and Implementation
This system’s primary consumers are daily commuters who, via their vehicles, demand road space and parking facilities. On the supply side of the system are local garage owners. Their garages may be either unused or made available to the public for parking services if the respective owners are willing to alter their resources to accommodate such facilities. The system has a third stakeholder, a private intermediary company that will own the system and controls the exchange between consumers and suppliers.
The applications developed and the parties communicating are a consumer-side mobile application and a web application. Alongside these applications, a RESTful Web Application Programming Interface is designed to wrap up this API-centric system through which both applications may communicate and access the system’s database simultaneously. We selected Microsoft Azure’s Software as a Service (SaaS) model as the cloud provider for all our applications. The Web API will hold all the necessary algorithms attributed to a successful PAYG smart parking system, including the all-important scheduling algorithm together with the quotation generation algorithm. The IDFP algorithm, on the other hand, will be situated on the front-end of the web application whenever a garage owner requests to add a new garage or edit an existing garage’s data. via the consumer-side mobile and web applications, commuters are provided with a user-friendly user interface allowing for efficient search functionality for available garages at any specific period. Passengers may also be informed about all the necessary information, including the garage pricing, garage features, and spatial data, which will help the user arrive at a particular garage. On the other hand, garage owners are the main drivers of the pricing charged to the consumers. Upon registering their resource, the owners will decide the whole pricing to be charged to the commuters allowing for a more competitive market. A detailed pricing portfolio may be specified, including
i) Rates specific to any length of a stay are 30 min, 1 h, 1.5 h, and other longer stays on any day of the week and time.
ii) Rates specific to any length of a stay are 30 min, 1 h, 1.5 h, and other longer stays on a particular day of the week at a specific time range.
However, the designed IDFP algorithm is providing the owners with a suggested price on each pricing category, described thoroughly in the next section. It will be at the discretion of the owner to abide by this suggested price to allow for a more competitive market. Owners are also entitled to specify daily garage availability, including night hours, and they may even temporarily close down the garage. The data stored in this system’s database is related to both the system’s business process and the IPA’s data. Due to the leading cloud computing benefits of scalability and availability and its developer-friendly architecture, we selected Azure Cloud Storage as the system’s database server location. We chose Microsoft SQL Server due to its harmonization with Microsoft Azure. The owner may decide his/her garage’s fees and not be specific to a particular date/time range.
5.2 Web Application Programming Interface
The system’s web application programming interface, now referred to as the Web API, will be the fulcrum of this API-centric system. The web and mobile applications will request already-processed data from this standard Web API, thus providing a smooth integration between the former two system components. Requests to the respective API endpoints will execute all the necessary queries and operations which constitute a successful PAYG smart parking system. Finally, this architecture allows the front-end applications to be as light as possible without re-processing the data. Instead of a Simple Object Access Protocol (SOAP) web service, we used REST (the representational state transfer) architecture as the API’s backbone. REST allows a uniform and consistent interface for CRUD (create, read, update, and delete) operations via intuitive Uniform Resource Identifiers (URIs) to the API’s endpoints. The web and mobile applications may receive and send a representation of all the necessary data—also referred to as resources-in either JavaScript Object Notation (JSON) or Extensible Markup Language (XML) format.
5.3 The Intelligent Demand-Forecasting Pricing (IDFP) Algorithm
The PAYG smart parking system’s pricing strategy is governed by an IDFP algorithm, shown in Fig. 1, primarily based on geospatial data. The algorithm’s outcome is an updated suggested fee per hour generated at run-time, allowing the system to recommend garage pricing portfolios to garage owners on garage registration and editing. As will be analyzed below, Google Maps JavaScript API [46] is consumed to achieve all the necessary geospatial data. We should note that the Google Maps services used have a standard free limit of 2500 elements—a combination of origin and destination coordinates-per day and a maximum of 100 elements per request [47]. A PAYG subscription will need to be purchased with a charge of $0.25 per 1000 elements if more elements are requested. Thus, it is beneficial to use Google Maps JavaScript API’s Geometry library [48]. This provides a means of limiting the cost of consuming the Google Maps JavaScript API. Hence, we ultimately decided to implement this algorithm using JavaScript and the supporting JQuery library. Earlier, we analyzed how existing pricing methodologies are used in parking solutions to tackle traffic congestion via the decrease in parking orbiting time. We did it by manipulating the parking fee to match the desired demand either at specific times of the day, as in SFPark’s demand-based pricing [9,28], or in general, as used in Crete’s contingent valuation methodology [42]. By considering the economic factors, we design an intelligent pricing algorithm that will forecast a particular garage parking demand at a specific date and time. This will lead to a suggested fee per hour provided to garage owners when setting up their desired price. The algorithm considers the following factors: (a) The proximity to the local leading/arterial road network, (b) The proximity to main commercial activity on the island and also the primary season specific to each, (c) The proximity to any on-going nearby events, (d) The garage’s condition or features, (e) Peak Hours.
Figure 1: Pseudocode of the intelligent demand-forecasting pricing (IDFP) algorithm
First and foremost, geospatial data about the local leading road network and local commercial sites is needed to compute a garage’s driving distance to any of these. We obtained the 126 roads’ list forming the Maltese leading road network from Transport Malta in July 2015 [49]. The starting and ending points of each road were manually noticed and inputted in the database. A JavaScript algorithm was designed, which used Google Maps Directions Service [50] to gain multiple coordinates along each main street in driving mode. The coordinates generated were then stored in the database through the respective Web API endpoint. We may apply the same to the coordinates of main commercial location areas. In this case, we may manually get a relatively small number of coordinates along the outskirts of these locations, forming a polygon enclosing the area. Provided with all coordinates/waypoints of each main street, several points along the border of each commercial area, and coordinates of nearby events, the algorithm may now compute a garage’s driving distance from the nearest main street, commercial area, and event.
5.5 Calculating the Proximity to the Main Road Network
Using the Distance Matrix service provided by Google Maps [47], a list of primary road coordinates as the origin and a garage coordinate as the destination is obtained to compute the garage’s driving distance from each street coordinate, and the nearest street coordinate to a garage. The system checks whether a garage is near a specific road according to a specified tolerance—a value indicating the degree of proximity through the Google Maps API. The closest driving distance from the nearest main road to the garage is estimated to get the respective proximity category and the underlying weight.
5.6 Calculating the Proximity to Commercial Areas
The algorithm will find a garage located in a commercial area at a particular time of the year; else, it estimates the walking distance from a garage to the nearest commercial area polygon coordinate using the Distance Matrix, the appropriate proximity, and the respective weight.
5.7 Finalizing the Suggested Fee
Provided with the garage’s proximity to the nearest main road and closest commercial area or event and the garage’s rating—and thus the weights attributed to each—we can compute the suggested fee. The base fee per hour is €1.69 per hour gained from the survey. The weights attributed to each demand factor of a specific garage are then put along this base fee-generating the suggested fee per hour.
We developed a scheduling algorithm, as shown in Fig. 2, that returns a list of garages available in a particular time slot.
Figure 2: Pseudocode representation of the scheduling algorithm
5.9 Fee Quotation Generation Algorithm
Whenever a user requests the fee to be charged on a booking in a specific time slot in a particular garage, the fee quotation generation algorithm, as shown in Fig. 3, will execute. This will return the exact total fee to be charged in this scenario, considering the booking duration, special fees specific to that time slot, and the duration spent in regular/normal time slots.
Figure 3: Pseudocode representation of the fee quotation generation algorithm
5.10 The Web and Mobile Applications
We developed the consumer-side web and mobile applications, as shown in Figs. 4–10, including the user interface (Figs. 4–9) and the owner interface (Fig. 10). The user interface facilitating users to primarily:
1. Search a list of available garages in a specific date/time range (Figs. 5, 6),
2. Filter garages according to a price range and/or its driving distance from the users,
3. View geospatial data of garages to shorten the arrival time to garages (Fig. 7) and View pricing information (Figs. 8, 9), mainly the fee for a specific booking date/time range
Figure 4: Web and mobile application home screen
Figure 5: List of garages in web and mobile application
Figure 6: List of garages on a map in web and mobile application
Figure 7: Selected garage on the map in web and mobile application
Figure 8: Selected garage’s description in web and mobile application
Figure 9: Garage price and booking screen in web and mobile application
Figure 10: Garage owner user interface for pricing portfolio detection in web application
Garage owners will use the web application to register and edit their garages, including the whole pricing portfolio, i.e., each pricing category’s suggested price. We used the ASP.NET Model-View-Controller framework as the web application’s back-end architectural pattern primarily due to it being industrially-recommended and fitting harmoniously into the system’s requirements. The developer-friendly concept of separation of concerns separates the data viewable by the user, processing and controlling, and actual structure of that data. In conjunction, we used HTML5, CSS3, JavaScript, and JQuery to provide the needed user interface alongside the industrially-used Bootstrap [51] due to its more efficient web development stage and easy-to-use mobile-first capabilities. We developed the mobile application natively for Android to provide a unique native feel to its user interface [52].
6 Testing, Evaluation, and Results
The scheduling (Tab. 5), fee quotation algorithms (Tab. 6), and demand-forecasting pricing (Tab. 7) algorithms were tested using manual testing rather than automated testing to ensure that all the results are conceptually appropriate according to the data provided. Once the algorithms successfully passed the manual testing, the user acceptance testing is carried out to evaluate the critical consumer-side processes, particularly the pricing, critical consumer-side processes, particularly the searching for garages and the booking process. In our evaluation of this system, we consider its usability a crucial feature of such a PAYG smart parking system. Multiple stakeholders interact with each other via a shared medium. A detailed user acceptance testing was carried out whereby, via numerous one-to-one interviews, respondents commented on various aspects of the system. The next part of our evaluation focuses on the analytical comments presented by an industry representative and a Transport Malta representative who critically analyzed the system’s main components, primarily on its development techniques and its potential for commercialization in the Maltese Islands. The participants mentioned that the navigation of the web application from the commuters’ side and the garage owners’ side is intuitive and easy-to-learn. Each webpage was designed in a user-centered approach to aid both commuters and garage owners in their individual needs. Moreover, the search and filter functionality for available garages at a specific date/time range and in a particular price range was considered to have a high degree of user-friendliness. One respondent even explicitly mentioned the easy flow from searching available garages to the final booking confirmation. After being presented with the web application, respondents highlighted the identical functionality available on such a smaller screen with a high usability degree. The API-centric approach chosen ensured the mobile application as efficient as possible. Apart from adequate navigation, they were again pleased with the flow of actions from the searching and filtering of available garages—which by default is at the current time-to the quick booking process. The system will work in a top-up-oriented payment process whereby users top up an amount every while, and credit is deducted on each booking. This, in turn, speeds up the secure booking process. Finally, participants analyzed the hourly rate presented in different scenarios, primarily based on a garage’s location. Respondents commonly agreed to all suggested hourly rates in different scenarios, which positively light the pricing weights generated in this paper. In conclusion, the participants were optimistic that such a system could be commercialized in Malta as they see it has vast potential. While as a result of this showing that the suggested pricing could give more weight to a garage’s proximity to commercial areas, the participants clearly showed that they would be willing to pay such hourly fees at the given scenarios via this PAYG smart parking system.
Table 5: Scheduling algorithm testing results
Table 6: Fee quotation generation algorithm testing results
Table 7: Demand-forecasting pricing algorithm test cases
Legend: P.C = Walking Distance to Commercial Areas in Metres; P.M = Driving Distance to Main Road Network in Metres; P.N.E = Walking Distance to any nearby events in Metres; G.R = Garage Rating; C.S = Current Season ID; P.H = Peak Hour? T = True F = False.
The scheduling algorithm holds an average response time of 169.2 ms, whereas that of the fee quotation generation algorithm stands at 126 ms. The demand-forecasting algorithm runs asynchronously while a garage owner adds or edits other details and returns a result within 12–18 s, which is acceptable considering its complexity. A user may add or edit further details in the meantime. Finally, analyzing the web application on a standard benchmark of GTmetrix [53], a PageSpeed and YSlow grades of A and B were achieved, respectively. The potential improvements are the use of a content delivery network (CDN) and expire headers. As the demand-forecasting algorithm lies within this web application’s script files, such scripts were not compressed, which would otherwise result in better performance.
With a nominal GDP per capita of €20212 and a 52 week year with 40 working hours, one minute is valued at €0.16. Our survey found that almost 74.4% of the respondents spend between 5 to 15 min cruising for parking during the week. Thus, with a confidence interval of 
6.14, it may be stated that a minimum of 68.3% of the Maltese driving population, i.e., 178274, will potentially save between €0.80–2.40 a day. Moreover, from our survey, it was found that, on average, the weekly cost of fuel is €21. With an estimated average traveling time of 90 min per day and the per-minute fuel cost of 0.03€, we should add a charge ranging from 0.15€to 0.45€to the above figures. This aggregate saving per user may result in a higher willingness to pay for these parking services, which, when also considering the potential 35400 everyday users and 46739 daily garage car parking spaces, increases this system’s potential.
This paper proposes a novel green IoT based PAYG smart parking system utilizing unused garage spaces at any time of the week. Using an ASP.NET Web API where all the main functionality for a successful PAYG system is programmatically stored, the client-side web and Android mobile applications successfully allow the necessary communication between garage owners and commuters. An IDFP algorithm was designed and implemented to suggest garage owners fees that better match a specific garage’s demand and location. We found the garage’s location as a leading determinant to the demand for garage parking. The evaluation of several fee scenarios shows that the algorithm’s design successfully forecasts the demand for a particular parking space at a specific time of day and year. Furthermore, these results continued to prove the initial hypothesis that a garage’s location is the leading determinant for such demand, particularly its proximity to a commercial area. The proposed IPA continues to be optimized to match demand better. The evaluation also proves that the potential users will be saving between €0.95 to €2.85 worth of orbiting cost per stay.
As future work, we recommend implementing artificial neural networks (ANN) tools that may automatically learn and hence update the various pricing weight categories stored in the system’s database. To facilitate the ANN learning process, it will use the algorithm’s suggested price, the actual price dictated by the owner at each price category, and each garage’s current proximity categories. We also recommend altering weights attributed particularly to a garage’s proximity to commercial areas to better match an increase in demand. Finally, one should consider commercialization issues other than the pricing mentioned above. Apart from further optimizations to the overall user experience, enhancements in the implemented algorithms’ efficiency, particularly the scheduling algorithm is recommended. By implementing time-blocks in the system’s database, the scheduling algorithm may efficiently scan the list of available garages when the system is successfully expanding. Finally, digital certificates can take care of security issues concerning the booking and payment process. Smart contracts implementation can ensure privacy to have a marketable and successful green IoT based PAYG smart parking system.
Furthermore, we can implement green communication [54] among the IoT devices and optimize power consumption using predictive models for battery life [55,56] and energy-efficient cloud-based Internet of Everything (EECloudIoE) architecture [57]. We can have the proposed smart parking system as a part of an Intelligent Transport Service (ITS) [58] in smart cities. As IoT devices collect and communicate data, it is crucial to ensure its safety [59] and reliability [60]. Due to the COVID-19 Pandemic, it is essential to appropriately track vehicles in the smart parking and ITS using contact tracing [61].
Acknowledgement: We thank all research participants Universities for their participation in the survey.
Funding Statement: This work is partially funding by the University of Malta’s Internal Research Grants.
Conflicts of Interest: The authors declare that they have no conflicts of interest to report regarding the present study.
1. C. Meilak, L. Garg, C. Attard and P. Xuereb. (2015). “Investigating different parking systems and issues,” in WCTRS SIG G3–Urban Transport Planning and Policy, Valletta, Malta, pp. 1314. [Google Scholar]
2. I. Aydin, M. Karakose and E. Karakose. (2017). “A navigation and reservation based smart parking platform using genetic optimization for smart cities,” in 2017 5th Int. Istanbul Smart Grid and Cities Congress and Fair, Istanbul, Turkey. IEEE, pp. 120–124. [Google Scholar]
3. T. J. B. Durga Devi, A. Subramani and V. K. Solanki. (2017). “Smart city: IOT based prototype for parking monitoring and management system commanded by mobile app,” Annals of Computer Science and Information Systems, vol. 10, pp. 341–343. [Google Scholar]
4. J. Rico, J. Sancho, B. Cendon and M. Camus. (2013). “Parking easier by using context information of a smart city: Enabling fast search and management of parking resources,” in 27th Int. Conf. on Advanced Information Networking and Applications Workshops, Barcelona, Spain, pp. 1380–1385. [Google Scholar]
5. D. Shoup. (2006). “Cruising for parking,” Transport Policy, vol. 13, no. 6, pp. 479–486. [Google Scholar]
6. N. Paramanatti, S. Naik and B. R. GR. (2020). “IOT based smart parking system for metropolitan cities,” Studies in Indian Place Names, vol. 40, no. 50, pp. 3722–3730. [Google Scholar]
7. A. A.Saleem, H. U. R. Siddiqui, R. Shafique, A. Haider and M. Ali. (2020). “A review on smart IOT based parking system,” in Int. Conf. on Soft Computing and Data Mining, Melaka, Malaysia, pp. 264–273. [Google Scholar]
8. Q. Long and X. L. Duan. (2020). “Optimization of road safety assessment algorithm for automatic driving vehicle,” 2019 Int. Conf. on Cloud Computing and Information Science, 27–29 December 2019, Shenyang, China, IOP Conf. Series: Materials Science and Engineering, vol. 750, no. 1, pp. 12173. [Google Scholar]
9. C. K. B. Krishnamurthy and N. S. Ngo. (2020). “The effects of smart-parking on transit and traffic: Evidence from SFpark,” Journal of Environmental Economics and Management, vol. 99, no. 6, pp. 102273. [Google Scholar]
10. W. Al Amiri, M. Baza, K. Banawan, M. Mahmoud, W. Alasmary et al. (2020). , “Towards secure smart parking system using blockchain technology,” in 2020 IEEE 17th Annual Consumer Communications & Networking Conf., Las Vegas, NV, USA, pp. 1–2. [Google Scholar]
11. S. Marusek. (2020). “Parking policy: The socio-legal architecture of parking bays in American cities,” Land Use Policy, vol. 91, no. 4, pp. 103931. [Google Scholar]
12. M. Giuliano, B. van Wee and T. Rye. (2015). “Urban parking policy in Europe: A conceptualization of past and possible future trends,” Transportation Research Part A: Policy and Practice, vol. 74, no. 6, pp. 268–281. [Google Scholar]
13. S. Singh and R. K. Tiwari. (2020). “Quality of service-based service selection in smart parking,” International Journal of Information and Decision Sciences, vol. 12, no. 2, pp. 154–175. [Google Scholar]
14. S. D. Nagowah, H. B. Sta and B. A. Gobin-Rahimbux. (2019). “An ontology for an IoT-enabled smart parking in a university campus,” in 2019 IEEE Int. Smart Cities Conf., Casablanca, Morocco, pp. 474–479. [Google Scholar]
15. N. Elkhalidi, F. Benabbou and S. A. E. L. Nawal. (2019). “Survey of reservation techniques in smart parking,” in 2019 1st Int. Conf. on Smart Systems and Data Science, Rabat, Morocco, pp. 1–6. [Google Scholar]
16. M. Sarangi, S. K. Das and K. S. Babu. (2019). “Smart parking system: Survey on sensors, technologies and applications,” in 2019 1st Int. Conf. on Advances in Information Technology, Chikmagalur, India, pp. 250–255. [Google Scholar]
17. T. Perković, P. Šolić, H. Zargariasl, D. Čoko and J. J. Rodrigues. (2020). “Smart parking sensors: State of the art and performance evaluation,” Journal of Cleaner Production, vol. 262, no. 6, pp. 121181. [Google Scholar]
18. S. Saharan, N. Kumar and S. Bawa. (2020). “An efficient smart parking pricing system for smart city environment: A machine-learning based approach,” Future Generation Computer Systems, vol. 106, no. 4, pp. 622–640. [Google Scholar]
19. S. Bhattacharya, S. Banerjee and C. Chakraborty. (2019). “Chapter 16: IoT-based smart transportation system under real-time environment,” in M. U. S. Khan, S. U. Khan, A. Y. Zomaya (Eds.) Big Data-Enabled Internet of Things, IET, IET Digital Library, pp. 353–372, . https://digital-library.theiet.org/content/books/10.1049/pbpc025e_ch16. [Google Scholar]
20. M. F. Afzaal, A. Rehman, S. Latif and N. A. Zafar. (2019). “Blockchain based automated formal model for safety and security in smart parking system,” in 2019 4th Int. Conf. on Emerging Trends in Engineering, Sciences and Technology, Karachi, Pakistan, pp. 1–6. [Google Scholar]
21. N. Buldakov, T. Khalilev, S. Distefano and M. Mazzara. (2020). “An open source solution for smart contract-based parking,” in Open Source Systems (OSS 2020IFIP Advances in Information and Communication Technology, vol. 582. Cham, Switzerland: Springer, pp. 55–69. [Google Scholar]
22. W. Al Amiri, M. Baza, K. Banawan, M. Mahmoud, W. Alasmary et al. (2020). , “Towards secure smart parking system using blockchain technology,” in 2020 IEEE 17th Annual Consumer Communications & Networking Conf., Las Vegas, NV, USA, 10–13 January, IEEE, pp. 1–2. [Google Scholar]
23. SpotHero. (2016). “SpotHeroon-demand parking marketplace,” . [Online]. Available: https://spothero.com/. [Google Scholar]
24. Paybyphone.com. (2016). “PayByPhone,” . [Online]. Available: https://www.paybyphone.com/how-it-works. [Google Scholar]
25. R. Borges and F. Sebé. (2019). “Parking tickets for privacy-preserving pay-by-phone parking,” in Proc.: 18th ACM Workshop on Privacy in the Electronic Society, London, UK, pp. 130–134. [Google Scholar]
26. Parkopedia. (2020). “Parkopedia-helping you park!,” . [Online]. Available: http://www.parkopedia.co.uk/. [Google Scholar]
27. Iparqcom. (2020). “iParq-intelligent parking,” . [Online]. Available: http://www.iparq.com/. [Google Scholar]
28. G. Pierce and D. Shoup. (2013). “Getting the prices right: An evaluation of pricing parking by demand in San Francisco,” Journal of the American Planning Association, vol. 79, no. 1, pp. 67–81. [Google Scholar]
29. Us.parkmobile.com. (2020). “Parkmobile | parking made simple,” . [Online]. Available: http://us.park-mobile.com/. [Google Scholar]
30. N. Farooqi, S. Alshehri, S. Nollily, L. Najmi and G. Alqurashi et al. (2019). “UParking: Developing a smart parking management system using the internet of things,” in 2019 Sixth HCT Information Technology Trends, Ras Al Khaimah, United Arab Emirates: IEEE, pp. 214–218. [Google Scholar]
31. Streetline, inc. (2019). “Parker (version 5.0.5.5) [mobile application software],” . [Online]. Available: https://www.theparkerapp.com/. [Google Scholar]
32. ParkWhiz and BestParking. (2020). “BestParking (version 7.5.4) [mobile application software],” . [Online]. Available: https://www.bestparking.com/. [Google Scholar]
33. Imperva. (2020). “ParkingPanda (version 4.4.0) [mobile application software],” . [Online]. Available: https://www.parkingpanda.com/. [Google Scholar]
34. MAWGIF. (2020). “mawgif (version 1.5.1) [mobile application software],” . [Online]. Available: https://www.mawgif.com/. [Google Scholar]
35. Municipality & Planning Department–Ajman. (2020). “Mawqifi (version 2.6) [mobile application software],” . [Online]. Available: https://play.google.com/store/apps/details?id=ae.gov.am.mawqifi.smartapp. [Google Scholar]
36. L. Benari. (2017). “ParkMe: A parking request app in Swift,” M.S. Thesis, Northridge, CA, USA, California State University, . [Online]. Available: http://scholarworks.csun.edu/bitstream/handle/10211.3/195231/Benari-Leor-thesis-2017.pdf?sequence=1. [Google Scholar]
37. M. S. Shiraj, M. Z. Shirgaonkar and M. S. Kumar. (2016). “Parkme: IoT based parking availability prediction in neural network,” International Journal of Science Technology & Engineering, vol. 2, no. 11, pp. 636–644. [Google Scholar]
38. JustPark. (2020). “JustPark,” . [Online]. Available: https://www.justpark.com/. [Google Scholar]
39. J. Rushi, N. Mehta and J. K. Bhaad. (2020). “An informative parking perspective: Justpark,” Studies in Indian Place Names, vol. 40, no. 51, pp. 86–90. [Google Scholar]
40. J. Sloman, A. Wride and D. Garratt. (2012). Economics. Harlow, England: Pearson. [Google Scholar]
41. Y. Geng and C. G. Cassandras. (2013). “New smart parking system based on resource allocation and reservations,” IEEE Transactions on Intelligent Transportation Systems, vol. 14, no. 3, pp. 1129–1139. [Google Scholar]
42. M. Anastasiadou, D. Dimitriou, A. Fredianakis, E. Lagoudakis, G. Traxanatzwe et al. (2009). , “Determining the parking fee using the contingent valuation methodology,” Journal of Urban Planning and Development, vol. 135, no. 3, pp. 116–124. [Google Scholar]
43. Cva.gov.mt. (2016). “CVA-controlled vehicular access,” . [Online]. Available: http://www.cva.gov.mt/. [Google Scholar]
44. National Statistics Office. (2009). “Malta. Lifestyle survey,” . [Online]. Available: https://nso.gov.mt/en/publicatons/Publications_by_Unit/Documents/01_Methodology_and_Research/LifeStyle_Survey_2007.pdf. [Google Scholar]
45. National Statistics Office. (2010). “Household budgetary survey 2008,” . [Online]. Available: https://nso.gov.mt/en/publicatons/Publications_by_Unit/Documents/01_Methodology_and_Research/HBS2008.pdf. [Google Scholar]
46. Google Developers. (2020). “Google maps JavaScript API v3 reference,” . [Online]. Available: https://developers.google.com/maps/documentation/javascript/3.exp/reference. [Google Scholar]
47. Google Developers. (2020). “Distance matrix service,” . [Online]. Available: https://developers.google.com/maps/documentation/javascript/distancematrix. [Google Scholar]
48. Google Developers. (2020). “Geometry library,” . [Online]. Available: https://developers.google.com/maps/documentation/javascript/geometry. [Google Scholar]
49. Transport Malta. (2020). “List of arterial & distributor roads,” . [Online]. Available: http://www.transport.gov.mt/admin/uploads/media-library/files/A%20%20D%20List%20042016.pdf. [Google Scholar]
50. Google Developers. (2020). “Directions service,” . [Online]. Available: https://developers.google.com/maps/documentation/javascript/directions#Directions. [Google Scholar]
51. D. Manuk. (2020). “Responsive catalog template,” Templateninja.net, . [Online]. Available: http://template-ninja.net/themes/dodolanmanuk/v1.1/black/index.html. [Google Scholar]
52. C. Chakraborty, B. Gupta and S. K. Ghosh. (2014). “Mobile metadata assisted community database of chronic wound,” Wound Medicine, vol. 6, no. 8, pp. 34–42. [Google Scholar]
53. GTmetrix. (2020). “How fast does your website load? find out with GTmetrix,” . [Online]. Available: https://gtmetrix.com/. [Google Scholar]
54. P. K. R. Maddikunta, T. R. Gadekallu, R. Kaluri, G. Srivastava and R. M. Parizic. (2020). “Green communication in IoT networks using a hybrid optimization algorithm,” Computer Communications, vol. 159, no. 3, pp. 97–107. [Google Scholar]
55. P. K. R. Maddikunta, G. Srivastava, T. R. Gadekallu, N. Deepa and P. Boopathym. (2020). “Predictive model for battery life in IoT networks,” IET Intelligent Transport Systems, vol. 14, no. 11, pp. 1388–1395. [Google Scholar]
56. S. R. K. Somayaji, M. Alazab, MK. M., A. Bucchiarone, C. L. Chowdhary et al. (2020). , “A framework for prediction and storage of battery life in IoT devices using DNN and blockchain,” . [Online]. Available: https://arxiv.org/pdf/2011.01473.pdf. [Google Scholar]
57. S. P. RM, S. Bhattacharya, P. K. R. Maddikunta, S. R. K. Somayaji, K. Lakshmanna et al. (2020). , “Load balancing of energy cloud using wind driven and firefly algorithms in internet of everything,” Journal of Parallel and Distributed Computing, vol. 142, no. C, pp. 16–26. [Google Scholar]
58. R. Kumar, S. Goel, V. Sharma, L. Garg, K. Srinivasan et al., “A multifaceted vigilare system for intelligent transportation services in smart cities,” IEEE Internet of Things Magazine, In press. [Google Scholar]
59. R. Vinayakumar, M. Alazab, S. Srinivasan, Q. V. Pham, S. K. Padannayil et al. (2020). , “A visualized botnet detection system based deep learning for the internet of things networks of smart cities,” IEEE Transactions on Industry Applications, vol. 56, no. 4, pp. 4436–4456. [Google Scholar]
60. M. T. Asif, N. Mitrovic, L. Garg, J. Dauwels and P. Jaillet. (2013). “Low-dimensional models for missing data imputation in road networks,” in 2013 IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vancouver, BC, Canada, pp. 3527–3531. [Google Scholar]
61. L. Garg, E. Chukwu, N. Nasser, C. Chakraborty and G. Garg. (2020). “Anonymity preserving IoT-based COVID-19 and other infectious disease contact tracing model,” IEEE Access, vol. 8, pp. 159402–159414. [Google Scholar]
 | This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. |