Abstract
The new IoT apps will not be able to inspire people to utilize them and may ultimately lose all their potential if an interoperable and trustworthy ecosystem is not provided. IoT has its extra security difficulties such as information storage, administration, privacy concerns, and authentication. The presently deployed IoT apps have encountered various security and privacy assaults globally. Due to being less secure and low powered, the IoT devices present a simple entryway to the adversaries to obtain access to the corporate networks, leading to giving easy control over all of the data of the users. The objective of this doctorate proposed work will be to solve the security associated difficulties in multiple IoT domains like the e-commerce, vehicular ad hoc networks (VANET), mobile ad hoc networks (MANET), and Internet of Drones (IoD). The proposed study focuses on the development of a distributed framework for IoT based on blockchain. The framework includes the usage of Ethereum-based smart contracts and auction models to increase the income and QoS for both the seller and the buyer and the development of a DAG chain-based distributed framework for parking lot allocation in a network of automobiles. The suggested model includes the requirement of obtaining agreement among the nodes with probability one in such a circumstance. The suggested model demonstrates to be predictable as typical voting-based consensus protocols like Practical Byzantine Fault Tolerance (PBFT) and at the same time can accommodate a high number of nodes even in an asynchronous setting. Research on Byzantine fault-tolerant systems has been ongoing for more than four decades, and although the solutions were shown to be feasible early on, they remained unworkable for a long time. With PBFT, the first feasible solution was provided in 1999, and this sparked fresh research that has resulted in unique applications that are still being developed today employing this technology. Despite the fact that the safety and liveness properties of PBFT-type protocols have been thoroughly investigated, when it comes to practical performance, only empirical results—often obtained in artificial environments—are known, and imperfections in the communication channels are not explicitly considered. It is our goal in this paper to propose the first performance model for PBFT that takes into account the effect of unreliable channels as well as the usage of alternative transport protocols across those channels. We also performed a large number of simulations to test the model and acquire a better understanding of the influence of different deployment factors on the total transaction timeframe.
1. Introduction
Our ability to link physical things all around the world to the Internet is developing at an alarmingly quick pace. Approximately 8.4 billion linked items will be present globally in 2020, according to Gartner research published last month [1]. According to the estimate, by 2022, the economy would have grown by up to 20.4 billion [2]. IoT applications are becoming more popular over the globe, with nations such as China, Western Europe, and North America [3] playing a big role in this development. After reaching 5.6 billion in 2016, the number of Internet-connected devices and communication ions between them is predicted to rise to 27000 million by 2024. The income produced by the Internet of Things business is predicted to expand from $ 892 billion to $4 trillion between 2018 and 2025 [4] according to forecasts. With such a substantial growth in the numbers, it is apparent that the Internet of Things is one of the most important forthcoming markets [5]. Figure 1 depicts the IoT architectures of the past, present, and future (prior, current, and future). It is shown in the image that there will be peer-to-peer connections between devices in the near future. Additionally, all of the gadgets will be linked to the internet as well. Another growing notion is that of the Social Internet of Things (SIoT), which may enable members of various social networks to be linked to gadgets [6]. There are some serious security and privacy problems that come along with the ever-increasing number of the Internet of Things apps that are being developed to make the lives of the general public more pleasant and easier. If an interoperable and trustworthy ecosystem is not accessible, the new Internet of Things apps will be unable to encourage people to utilize them, and they may ultimately lose all of their promise. The Internet of Things (IoT) introduces new security challenges, such as data storage and management, privacy concerns, and authentication. The Internet of Things (IoT) apps that are now in use have been subjected to a number of security and privacy breaches throughout the globe. Because they are less secure and low-powered, the Internet of Things devices serve as a simple route for attackers to get access to corporate networks, allowing them to obtain complete control over all of the data of the users in the process. In the last quarter of 2016, the Mirai assault 1 affected around 2.5 million devices, and DDoS attacks were conducted against those devices [7].
Patients have also benefited from the effective placement of the Internet of Things devices in their bodies to monitor a variety of conditions, such as heart disease [8], which has been shown in many studies. As a result, such Internet of Things apps might enable attackers to follow the movements of a specific individual, particularly if the person is a well-known or important figure. The compromise of these devices, despite the fact that there have been no real-world cases, may be very harmful. Cyber-Physical Systems (CPS) is another sector that may benefit from the rise of the Internet of Things (CPS). When using CPS, the items or objects that are present at a location are linked to the Internet through sensors, and actions are conducted in response to physical changes. Because CPS is connected with key applications (for example, automobile networks, mobile networks, and the Internet of Drones), it has major implications for security vulnerabilities. Four levels are critical in any IoT environment or ecosystem, and they are of the highest significance. For example, different actuators and sensors are utilized to perceive information, and data is added for various functions to be performed in the first layer. The second layer, which operates on the foundation of the first layer, transfers the data acquired via the use of a communication network. The third layer, sometimes known as the middleware layer, will be present in future Internet of Things applications. It serves as a connecting thread between the application and the network layer, allowing data to flow between them. The last layer, which is comprised of applications such as smart houses, smart meters, and smart automobiles, is the final layer. In addition to the four levels, there are other distinct gateways.
Smart contracts are basically computer programs that are recorded on a blockchain and are activated when certain criteria are satisfied. A common use of this technology is to automate the implementation of an agreement so that all players may be confident of the conclusion instantly, without the participation of a middleman or the waste of valuable time.
Hybrid blockchain technology is one of the several forms of blockchain technology available. Hybrid blockchain, on the other hand, is the last form of blockchain that we will explore in this article. A hybrid blockchain is a kind of blockchain that sounds similar to a consortium blockchain, but it is not one. However, there may be some parallels between the two of them.
Generally speaking, a hybrid blockchain is a mix of both a private and a public blockchain. It has applications in organizations that do not want to install either a private blockchain or a public blockchain but instead want to deploy the best of both worlds at the same time.
1.1. Types of Blockchain
An individual user does not need permission to become a miner or to participate in the blockchain network while using a permissionless blockchain, and anybody may join or leave the network at any time. The Bitcoin blockchain is the ultimate example of a permissionless blockchain. Even with a low transaction throughput, permissionless blockchains have the capacity to accommodate a large number of nodes in the network, allowing them to scale exponentially. In contrast, in order to join in the blockchain network in the permissioned blockchains, it is necessary to adhere to a set of regulations that must be observed. Comparing permissioned blockchains to permissionless blockchains, the notion of permissioned blockchains significantly enhances the total throughput of transactions done by users.
Private blockchains are very fast. This is due to the fact that there are less participants on the private blockchain compared to the public blockchain. Overall, less time is required for the network to establish consensus, which results in quicker transaction processing.
Private blockchains are more scalable than public blockchains. The scalability is made feasible by the fact that only a small number of nodes are permitted to verify transactions on a private blockchain. This implies that it makes no difference how large the network becomes; the private blockchain will continue to operate at its former speed and efficiency. The centralization of decision-making is the most important factor to consider here.
There are several benefits to using blockchain technology into the Internet of Things applications. Section 1 examines the numerous security problems that IoT applications have to contend with. Table 1 and Table 2 contains an overview of a few particular Internet of Things security concerns, as well as prospective blockchain-based solutions to each of these challenges. Some of the most notable advantages of using blockchain technology into IoT applications are as follows:
In the present day, computer systems are built on bits that can only be in one of the two states, namely, 0, or one of the two states, namely, 1, at any one moment [9]. Qubits, on the other hand, may be in both states (0 and 1) at the same time [10]; hence, this does not hold true. Superposition is the term used to describe this. Because of the superposition principle, quantum computers are capable of solving very complicated and difficult mathematical problems. This might be very beneficial, but at the same time, these quantum systems have the potential to compromise the security of systems that rely on the intricacy and difficulty of mathematical problems. It also implies that cryptographic methods may defeat hash and symmetric algorithms, which are predicated on the infeasibility of brute force assaults, as well as other algorithms. However, with the assistance of an exponential increase in calculation time, these algorithms are now capable of conducting a comprehensive search for secret keys [11]. Quantum information processing is greatly aided by the principle of quantum entanglement, which has been dubbed as one of the reasons quantum algorithms such as Shor and Grover [12] perform significantly better than classical algorithms and result in an almost exponential increase in computation times. Because of quantum technology, it is now possible for users to break the basic cryptographic algorithms that are the building blocks of the general blockchain-based algorithms that are used today. If a large-scale quantum computer is developed, the public-key cryptosystems would be compromised, rendering all of the data stored on the Internet and elsewhere accessible to unauthorized access, hence reducing the integrity of any form of digital communication [13]. In today’s world, the majority of the various blockchain-based algorithms utilized in various DApps are based on three cryptographic algorithms: the discrete logarithm problem, the integer factorization problem, and the elliptic-curve discrete logarithm problem. All of these issues, which would ordinarily be difficult to address, may be quickly solved with quantum computers [14]. Using fundamental techniques based on quantum computers, such as Shor’s algorithms, it is possible to quickly crack these difficult issues. In addition to Grover’s method, which is used to break symmetric key cryptography schemes that utilize the same key for both decryption and encryption, quantum computers also employ a number of other efficient algorithms. With the use of Shor’s and Grover’s algorithms [15], even one-way encryption schemes such as hash functions are at risk of being compromised.
A driving force behind the effort is the need to have a system that assists individuals in choosing the best parking place for their automobiles prior to them actually arriving at their final destination. They save a significant amount of time and energy since individuals are no longer needed to search for parking spaces. Already, a space has been assigned to them, and they have agreed on a suitable fee for that spot. According to the INRIX 2018 Global Traffic Scorecard (GTS), a motorist in the United Kingdom (UK) wastes an average of 44 hours per year trying to find a parking place [16]. In response to the irritation and frustration they experience, people get involved in conflicts and quarrels with other vehicles over parking places. The figures shown below clearly demonstrate that the rise in the number of automobiles on the road is directly proportionate to the issues that people will encounter in terms of time, health, money, social interactions, and their surrounding environment, among other things. As a result, it is critical to modernize the underlying design of the parking system and shape it in such a manner that it is dispersed, egalitarian, accessible, safe, and cost-effective for all users.
2. Related Works
This section discusses the algorithms, as well as their benefits and disadvantages, that are now being used in the construction of new parking lots as well as the redesign and distribution of existing parking lots. Using a vehicular ad hoc network (VANET), researchers Lu and colleagues [17] demonstrate vehicular transmission, which is utilized to search for vacant parking spaces in a vast territory. This research is funded by the National Science Foundation. At the time of the encounter, the automobiles speak with one another and exchange information about an available parking space in the nearby. Under this method, different users get information about the same parking place, and the space is allotted according to the first come, first served principle. According to the authors of [18], their technique includes an agent that is in charge of monitoring traffic data and assigning parking spaces to different cars in the most effective way. It is possible to analyze a substantial amount of the Internet of Things data, which is gathered via the various Internet of Things sensors put in parking lots, and to provide users with 39 parking information using a variety of techniques. It has been recommended by the authors of [19] that approaches be developed to analyze the data at the sensor level in order to prevent data from being presented inaccurately due to a computational delay in response time. Directing them to less crowded areas in their near vicinity would help to alleviate unnecessary traffic congestion for the other automobiles on the road as well. Specifically, as detailed in [20], a reservation-based system for parking a self-driven automobile has been proposed that eliminates reservation overlap while also addressing privacy and security issues while also ensuring reservation privacy and security. In their study, Shao et al. [21] discuss effective ways for employing sensor data to address parking infringement problems that have been validated in the field. The sensors collect information on cars that are in violation of parking laws and report it to the proper authorities for further investigation and enforcement. Because of the dynamic character of the vehicular network, the predictive algorithms are incompatible with the dynamic nature of the vehicular network as a whole. Afterwards, we will go through the most significant contributions that have been made in order to enhance the scheduling and distribution of parking spaces. A great deal of work has been spent into bringing the notion of a smart parking system to fruition, but the overwhelming majority of them are centralized in nature and depend on predictive analysis or data collected via the use of sensors placed in parking garages. It is also worth highlighting that tiny garage owners and open parking zones are still unable to register with the system since there is currently no such option. Therefore, a distributed architecture for the smart parking system is necessary that does not violate the privacy of users; is precise, efficient, and fast to respond to requests for parking spaces; and offers parking spaces at the most reasonable pricing possible.
It has been observed that the number of motor vehicles has increased at an exponential rate during the previous few decades. In conjunction with the rapid rise in the number of vehicles on the road, there has also been an increase in the demand for parking space. In densely populated areas, finding a parking space in a severely congested area may be a challenging task, especially when trying to park legally. At peak periods, drivers are sometimes forced to circle the parking lot many times before they can locate a single open parking place. Leaving your automobile in an illegal parking lot may result in a range of legal and security ramifications, including a traffic ticket, towing, and theft or burglary, among other things. When travelling by public transit, people often forego their comfort in order to avoid dealing with parking hassles on their destination. Individuals spend a substantial amount of money in parking lots that are governed by a variety of centralized organizations in order to get a parking place that is suitable for their purposes. In spite of the fact that a range of activities have been taken to address the problem of parking space distribution, the majority of these efforts have been driven by forecasts based on historical data. In addition to increasing with each passing day, the number of motor vehicles is increasing, and this is resulting in an increase in the dynamic nature of the vehicular network. As a result, the prediction models that are now in use are unable to provide trustworthy findings, and the situation becomes even more serious. In recent years, a number of Internet of Things (IoT) applications have been developed to guide automobiles toward a parking place appropriate for stopping. On the other hand, since they need many infrastructure upgrades, the collection of appropriate data, the deployment of sensors, the payment of installation and maintenance charges, and so on, they constitute a major financial investment. Furthermore, the reach of these apps is limited to the particular parking premise in which they are installed and maintained. Therefore, although this solution is completely adequate when restricted to a certain parking zone, it does not have a feature that enables the user to assign parking from start to end during the whole process. The centralized structure of parking lot allocation prevents us from making small garages or empty parking zones accessible to users, and the incapacity of users to successfully register with the program means we are unable to make empty parking zones or tiny garages available to users. This chapter introduces an original concept for a distributed platform, which may assist anybody in making money by simply registering a suitable parking space that can be used by motor vehicles. This concept is described in detail later in the chapter. It is made possible for users to hunt for a parking place that is accessible in their close vicinity in order to avoid travelling unnecessary miles. It will become clear over a period of time that the system is both cost-effective and beneficial to both the parking lot’s owners and the customers that use it. The use of this strategy will prohibit the owners of big parking lots from collecting a hefty cost for the privilege of providing a parking place for automobiles.
3. The Work That Is Being Considered
There is a significant reliance on the blind search approach in the present parking allocation algorithm, in which the end user is given with information on a large number of parking spaces in their neighborhood, but after that, the driver is left to fend for themselves. These parking scheduling systems are mostly reliant on centralized private firms, who are renowned for burning a hole in their clients’ wallets.
When it comes to cloud data storage, there are two types of failures that stand out: Byzantine- and crash-related problems. Both of these terms refer to systems that either perform well or do not react at all after experiencing an initial failure. As opposed to this, Byzantine faults allow for arbitrary failures and, as a result, do not restrict the capabilities of an attacker. As a result, they are well-suited for our methodology, which models cloud storage of sensitive data as being susceptible to random defects, such as malevolent attackers, network outages, or memory corruption, among other things.
An often-used protocol in the Byzantine context is the practical BFT (PBFT), which includes the versions Zyzzyva [7] and Aardvark [4], as well as their derivatives. It is a leader-based system that relies on majority voting among all participating servers, as well as strong encryption, to ensure message ordering and robust consistency in the face of Byzantine faults and other problems. It is necessary to have operational servers with communication connections between them in order to allow for majority voting. The enhancements of [9] demonstrate how privacy may be linked with secret sharing if it is required in addition to secret sharing.
In most cases, two kinds of deployments or applications may be distinguished: local area networks (LANs) and blockchains. When installed in a closed network inside a single administrative domain, i.e., in a LAN, such as the “5 Chubby nodes within Google” scenario, the use of UDP for message transmission results in the greatest performance. Every PBFT transaction is connected with a “view” that has a dedicated primary in order to ensure that the transaction remains active in the event of an error or a transaction that does not continue. An automatic view change is triggered if a transaction fails to make progress or when a leader is suspected as being malicious by the nodes. This is done in order to avoid potentially problematic primary from slowing transaction processing. View changes occur when a new leader is chosen and operation proceeds under the direction of this new leader. For consistency considerations, it is assured that transactions that have already been committed in the old view will be appropriately handled by nodes in the new view. However, as the studies of [3] shown, packet loss may occur even in the most optimal LAN configuration owing to congestion, and the resulting triggered view changes have a detrimental effect on performance.
Different assumptions and conditions apply in the blockchain world [8, 11, 16], and findings cannot be simply transferred from one world to another. Many transactions are often batched together, and consensus is grouped into epochs that include all presently outstanding transactions in a certain period of time. Furthermore, transaction durations are often expressed as amortized values, which makes sense in the context of a blockchain with a constant incoming stream of transactions and a sufficient number of buffered transactions in each epoch.
Assumptions made in the models include the fact that a reliable channel can always be formed with negligible overhead over faulty channels and that the network buffers at nodes are limitless. In reality, if authenticity is necessary, they often use TCP or its secure variation TLS as the transport protocol.
The term “benchmarking” refers to the process of comparing and estimating the performance of different BFT protocols (5, which is used to compare and evaluate the performance of different protocols). Stochastic reward nets (SRN) are used to describe the “mean time to full consensus” in [14], which is the only known more systematic technique.
Individual distributions from measurements are used to model the network as a trustworthy channel in which the rate of message transmission between all pairs of peers is the same, and the model is fitted to the data.
For the sake of summary, while there is a large body of research in BFT and many protocols have been proposed and benchmarked, there is little knowledge about performance modeling of such protocols, specifically considering the impact of the underlying communication channels, such as packet loss or latency caused by the physical channel characteristics, which is a major source of confusion. A further limitation is that only a small number of findings are backed by free software implementations, making verification and comparison of results difficult or impossible in many cases..
3.1. Distributed Parking Lot Scheduling
Because of the centralized character of the plan, open spaces and tiny private garages cannot be made accessible to the general public for parking purposes according to the nature of the plan. The parking scheduling system makes it possible for individuals to publish their free/vacant spots, which can subsequently be utilized by other users to park by using a distributed framework algorithm for the parking scheduling system. Using a directed acyclic graph (DAG), the motorist may arrange parking by surrendering a transaction, and the owner of the parking space can subsequently recommend parking spaces that are accessible to the user. Between the user and the owner, a smart agent is installed in order to filter out unnecessary information.
3.2. Digital Identity
The private key is a big collection of random numbers generated by a random number generator and stored in a secure location. Every transaction that takes place between the nodes of the network makes use of this set of keys in order to avoid the problem of nonrepudiation and to guarantee that there is no confusion about the allocation of resources.
After a user’s keys have been created, he or she may request that a parking place be assigned to them. Before being assigned a slot, the user is required to fill out a number of forms, including information about the location, the time and length of the slot need, the maximum price the user is prepared to provide, and so on [21]. This information is then packaged into a transaction container and sent to the various nodes in the network. The event architecture is shown in Figure 2. This occurs in the form of a gossip event because the gossip protocol needs far less bandwidth to gossip the DAG that has been established, and when the users “gossip the gossip,” the reach of the network increases greatly. The next part focuses on the scheduling and distribution of parking places in accordance with the gossip protocol. Event is synonymous with block in generic blockchain, and the request for a parking space booking is synonymous with transaction in blockchain, respectively [22]. Figure 3 and Figure 4 illustrates how each event or each block might have numerous requests for slot bookings in the same time frame. There are many parking slot requests associated with each event in the graph. These requests include the current parking slot request made by the user as well as some prior requests that have not yet achieved a consensus and are now flowing via the gossip protocol.
3.3. Consensus Algorithm
Due to the fact that there is no centralized body in a distributed system that can bring all the nodes to an agreement, some frequently used consensus algorithms, such as Raft and Paxos, establish agreement by designating a leader to the system. However, when it comes to denial-of-service attacks and interminable delays, the consensus algorithms are practically defenseless. Even a single node may be compromised, culminating in the infiltration of the whole network as a consequence.
When it comes to open distribution networks, the POW (proof-of-work) approach is used to tame the aforementioned challenges and reach an agreement on how to resolve them. The POW method allows the node that answers a random mathematical puzzle in the least amount of time to combine the blocks in order to do the addition in the smallest amount of time. In order to make effective use of the generic POW algorithm, it is necessary to keep a few things in mind. There are two issues with this kind of thinking. As a consequence of the addition of extra delay caused by the addition of a block, there is an increase in total power consumption of the approach, as seen in Figure 1. For the second time, since a transaction or communication that has been included in a block may be removed at a later time, this approach does not guarantee that the block will be decisive. The POW method, on the other hand, fails to maintain a properly ordered record of the communications since there is no correlation demonstrating a link between the sequence in which transactions enter a network and the sequence in which they are conducted in the algorithm. Due to the fact that the POW algorithm is largely employed in Bitcoin, it may seem that keeping an ordered record of transactions is not important. The capacity to retain an ordered record is crucial when dealing with a smart parking scheduling and allocation system, since the number of users and the number of available parking spaces is not inversely related. So it is critical to examine the time of slot booking and then distribute the slots among users in a way that is as effective as possible.
The POET (proof-of-expired-time) algorithm is shown in this work as a technique of realizing the efficiency and competence of POW without the requirement for extra electrical overhead to achieve these goals. Despite the fact that this strategy reduces waste of resources, it does so at the expense of the trust factor in the organization. Attempting to address this issue, the Byzantine agreement system incorporates the votes of the members into its decision-making process.
It is recommended that a DAG-based agreement technique be utilized to solve the aforementioned challenges, which are intrinsic to the generic consensus process and which are discussed in this study, in order to overcome them. It has been shown that, while giving all of the advantages of blockchain, this technique is efficient, quick, and predictable, and it also prevents any legitimate block from becoming a fork at the same time. By using this approach, which is based on the notion of virtual voting, it is possible to reduce the requirement for excessive overhead messages.
3.4. Allocation of Slots
The popularity factor has a significant impact on whether or not a witness receives a good or negative response. Everyone now participates in virtual voting, where each node is capable of secretly determining how popular each witness is among all of the nodes and subsequently casting a vote for the witness in question. This facilitates the avoidance of the real voting procedure [23]. Byzantine nodes are also taken care of since virtual voting is handled under uncompromising and stringent criteria, and the nodes are given the authority to calculate the votes for other members on their behalf. If any of the nodes’ witnesses in round have an association with witness w +1, then witness w +1 gets a positive vote, while the nodes having an affiliation with witness w +1 get a negative vote. W +1 is now deemed a popular witness if it receives a favorable vote from two-thirds of the witnesses in the round; otherwise, it is voted as a nonpopular witness by the remaining witnesses. This assures that, despite the fact that the nodes are independently calculating and assessing the votes, they all have the same consistent copy of the distributed application graph. As a consequence, there is a 100 percent byzantine agreement since all of the votes are treated the same.
3.5. Request from the Input Source
It is possible to get a general agreement with probability 1 once we have determined the precise sequence of the occurrences and after we have settled on the most popular witnesses. During a round, it is necessary to compute a time-stamp for every event since any event may be associated to a transaction, regardless of its popularity or the character of the member, i.e., whether it is a witness or not, and hence, every event must be timestamped. It is determined by analyzing the earliest timestamp of the popular witnesses who have come into contact with the transaction and calculating the median of all those timestamps together. In this case, the median value is the value that is closest to the center of all of the timestamps and that stays constant or unaffected by extreme values [24]. This median timestamp corresponds to the agreement timestamp for this specific occurrence. This is done for all of the events on the graph and for each one of them. Following that, the events are ordered in the sequence in which they will occur during transaction processing. It is preferred to calculate the median rather than the mean in order to prevent the discrepancy that would occur if specific timestamps were at extreme values (i.e., far from the middle value), since it would be ignored and the median value will remain unaltered.
3.6. Implementation of Proposed Work
The network model that has been built caters to a wide range of users, including drivers, parking space owners, small and large parking lot owners, and other individuals and organizations.
Users may register as members of the network after providing some basic information, which will allow them to plan and reserve a parking place in advance. Specifically, the user is asked for information regarding the stopping place or region, the time of arrival, and the amount of time they want to spend at the location. The latitude and longitude of the user’s port of call are marked by the letters dlat and dlon, respectively, while Edur indicates the length of time the user will need the parking space and the time of arrival, and Edur signifies the length of time the user will require the parking space. As a precaution against the possibility of unused reserved parking spaces being available, the user will be granted a slot while being required to adhere to a time constraint in order to prevent this situation. If, on the other hand, the user fails to attend within 30 minutes of the time slot they have scheduled, their reservation will be cancelled, and the seat will be given to someone else. This ensures that clients do not book parking spaces that are not essential and that they reserve a parking spot in a rational manner after picking the most suitable time slot. The adoption of the DAG technique provides the user with the certainty that they will absolutely get the time slot that they have requested, regardless of how complicated the request is. The user is less inclined to make further bookings since they are certain that they will get a spot. It increases the efficiency of the operation for selecting a time slot. As soon as a choice has been made about a parking space, the user is presented with the final optimized parking slot or spaces. In order to prevent spam from being sent to the user, the PL owner will be unable to send spam to the user. Regardless of the number of recommendations sent to the system, it only chooses those that are adequate and meet the requirements set by the users in the first place. On the other hand, some people are not bothered by paying a penny more for a parking spot that is closer to their destination yet costs less but is located a little farther away from their final destination. In the event that they were worried about the safety of their vehicle, some individuals could prefer a congested parking lot. It is possible that changing the values of these weights will have an influence on the optimal solution that is shown to the user.
The cost of parking spots is not predetermined under this technique. In a particular place, prices are decided in line with the permissible range defined by government authorities in that location. Despite the fact that there are several factors that determine price, the following are the most significant. These factors include the availability of parking spaces, the necessity for parking spaces at a certain time, the anticipated amount of time spent pulling over, the day and time of the week, and whether it is a festival or a typical day. It is possible to discover the optimal price by using the dynamic pricing algorithm, which will be explained in further depth later. The dynamic pricing algorithm takes into consideration all of the features described above, as well as a number of additional parameters, before calculating a price to charge. Using the Hungarian technique, it is possible to identify which parking spot is the most suited for a certain user after calculating the prices of various parking spaces in a given location. The Hungarian technique has the advantage of being able to resolve the assignment problem in polynomial time, which is a substantial benefit. The program determines the lowest possible price for each user; therefore, this strategy assures that each user gets just one and only one spot at the lowest possible price.
3.7. Dynamic Pricing
In this section, you will learn about the process that is utilized to calculate and propose a dynamic pricing for parking spots. We will assume that a slot has a 24-hour life cycle and that pricing will need to be determined on an hourly basis. Under the dynamic pricing approach, the income production of all PL owners is boosted, while the cost of parking for individual users is optimized. If the final price is , it must fall within the range of the maximum and lowest base prices.
We believe that every user would like to park his or her car as near as possible to their intended location, and thus we assume that this is the case. Even if the consumer had to pay a somewhat higher price than the standard fee to secure a slot, the proximity to the ultimate destination would frequently be a significant advantage [25]. As an alternative, if the costs are at the lower end of the spectrum, i.e., nearer to the least base price, the chances of locating parking spaces a little further away from the destination may be increased. The suggested strategy is successful in gaining support from both users and PL owners in the neighborhood area since it differentiates parking prices based on their proximity to one other. where is the inflated price due to distance parameter and
Equation 4 represents the detection of slots in the network region.
It is reasonable to assume that the PL owners would want to maximize their earnings and that there would be fewer unoccupied slots available as a result. As a result of using this strategy, if the number of available parking spaces in one specific place exceeds a set percentage of the total number of available parking spaces, the PL owners are barred from raising the price of parking spots in that region. However, if the number of available parking spaces in an area is fewer than a certain proportion of the total number of available parking spaces, the PL owners may tolerate an increase in price by a unit value while still managing to fill the majority of the available parking spaces.
The amount of variance in the price of parking spot allocation is influenced mostly by the density of traffic. An hourly examination of the road traffic has yielded these data. The following are the elements that have the most influence on the pricing fluctuation caused by traffic density: (1) the time of day when the parking space is reserved, (2) the day of the week on which the parking slot will be reserved, and (3) the date of the year on which the parking spot will be reserved.
Based on the analysis of these characteristics, the traffic density is calculated, and the costs may be raised or lowered as a consequence of the findings. After 8 p.m., the traffic density begins to decline and stays low until the wee hours of the morning. As a result of these considerations, parking rates may change accordingly.
Let the inflated price due to traffic density be represented by . (i)Normal traffic density scenario(iii)High traffic density scenario
The density of traffic is influenced not only by the time of day, but also by the day of the week, according to the study. So, the PL owners may manage to raise the rates during the weekdays while simultaneously lowering them over the weekends, as seen below. As a result, during the festival period, there is a larger than usual volume of traffic. Let the inflated price caused by the day and date parameters be denoted by the symbol . The rates may be changed if it is a weekday during festival season and they are within the permitted range.
3.8. Input Allocation of Price Range
There is one constant parameter that is shared by all PL owners, and that is the period of the reservation of the parking spot. As an incentive to clients, several local parking garages offer free first-hour parking in exchange for a hefty fee for any additional time after the first hour. Using the price change model, it is suggested that a change in the price of parking occurs every 8 hours of parking duration. For the first 8 hours of parking, the costs remain unchanged, indicating a pricing parity. Prices are raised by a constant amount for parking spaces booked for 8 to 16 hours and by double the constant amount if the parking spot is reserved for 24 hours.
The price inflation for parking durations ranging from 8 to 16 hours and 16 to 24 hours is shown in equations (11) and (12), respectively. After all of the price variation factors have been taken into account, the prices are then modified accordingly. The Hungarian assignment model is used to map each user to the most appropriate PL in the most cost-effective way at the end of the process.
4. Experimental Results
Traffic data collected from five different locations around Chennai is being analyzed in order to confirm the efficiency and performance gained through the use of the model under consideration. Table 3 displays the latitude and longitude coordinates of the selected place in the form of latitude and longitude. The amount of parking spaces available in each lot varies from one site to the next. Users may park in any open spot in any of the parking lots that are currently available.
A comparison of the proposed model with the greedy parking slot allocation as well as the fixed price parking slot allocation model is shown in Figure 5.
When compared to the fixed price and greedy models, the findings in the picture demonstrate some distinct behavior for various consumers seeking parking places. Figure 6 represents the fixed price and greedy models; the suggested model charges higher fees for the first two consumers, but the proposed model charges somewhat lower rates for the following two clients. Rather than assuming that parking lot owners in the greedy model are fully aware of the current traffic flow pattern, which we are attempting to determine the pricing for parking spaces based on, we are assuming that the parking lot owners in the greedy model are fully aware of the current traffic flow pattern, based on which we are attempting to determine the pricing for parking spaces. Because there is no consensus process to help in the order of requests that are constantly entering, and because there is no peer-to-peer network in the first model, it is the second model that is superior. On festivals and weekdays, all parking lot owners report a consistent increase in the prices they charge for parking spots, with no distinction made based on the number of requests received or the distance the customer travels to the parking space. There is a lack of total understanding among parking lot owners of the whole picture of parking requests coming in from various geographic areas. A rise in the price of free parking places is made without considering the whole picture, resulting in confusion and traffic bottlenecks, as well as a reduction in total income for the property owners. When comparing the suggested model to the greedy model, the total prices offered in the proposed model, which takes into account overall income as well, are often lower than the prices given in the greedy model. As a result of cooperation under this approach, the user’s experience is improved, and the income earned is also boosted.
Figure 7 depicts the rates of resource usage of parking spaces when the proposed model is not in use as opposed to the rates of resource utilization of parking spaces when the proposed model is in use, respectively. When the model is not in use, users have a tendency to park their cars in the largest available vacant spots that are visible to them, resulting in a poorly managed scenario characterized by traffic bottlenecks and turmoil, as well as parking lots that are not utilized even during peak working hours. Even after paying a premium charge, consumers are forced to drive around enormous parking lots in quest of an available parking space notwithstanding their efforts. A benefit of the suggested model is that it facilitates this process by providing a vacant slot in response to the user’s requests, and it does so at a lower cost in the majority of circumstances.
The customers benefit from the finest parking spots available at the greatest pricing, and the lot owners benefit from the optimal use of their spaces as well. As shown in Figure 8, the largest parking lot has a comparable use of its spaces, but the four other parking lots have a considerable improvement in utilization of their spaces. The increase in resource utilization would go a long way toward enhancing the incentive of small parking lot owners and private parking space owners to provide parking space for customers. This will, in the long run, improve the number of parking alternatives available to consumers while simultaneously decreasing the likelihood of a traffic bottleneck. What these two numbers show is that the .suggested model has increased the resource utilization of small parking lots while not interfering with the resource usage of larger parking lots that were already open to all customers. This obviously implies that some of the users were having a great deal of trouble locating a parking space in the larger parking lots since they were not aware of the smaller parking lots in their immediate proximity. As a result, the suggested approach appears to be superior in terms of both the general public and the owners of parking lots. Figure 9 depicts a timegraph containing the entrance and exit values of transactions utilizing a hash graph and a blockchain, respectively. On the one hand, we can observe that the hash graph maintains the order or transactions in terms of entrance and exit, but the general blockchain does not retain the order or transactions.
In general, it is preferable to utilize UDP rather than TCP whenever feasible, since TCP causes unacceptable performance deterioration when error rates on the transmission channel are greater than the tolerable threshold. Though adding nodes to a PBFT system is not necessary from a robustness standpoint, it turns out to be a good way to boost redundancy at the network layer when dealing with inconsistent communication. Additional benefits of using repetition codes include the fact that UDP may be used instead of TCP in cases when packet loss is high, which can result in large performance increases.
If the behavior of the channel is known in advance, we propose that the deployment be configured appropriately in order to remain in the UDP regime. At the end of the day, for our sort of application, a specialized network protocol that adaptively optimizes retransmissions and other parameters while not raising latency would be preferable.
It was discovered from the structure of the communication pattern that unreliable channels had varying effects at various stages of the communication pattern’s lifecycle. A node that does not get a single PRE-PREPARE message may already be out of sync with the current transaction; on the other hand, if f PREPARE messages do not come, the node will still have sufficient information to continue. This demonstrates that the initial broadcast from the main is disproportionately more significant than the rest of the messages, and that efforts taken to boost the chance of its success will have a disproportionately large influence on the success of the whole transaction. As a result, it could make sense to restrict the usage of TCP to this phase or, as we have done, to proactively repeat this message once or twice.
4.1. The Byzantine Case
If f nodes are really evil, their messages will be rejected by the honest nodes if they do not adhere to the protocol in their communications. In order to slow down transactions (and, thus, service time), the best they can do is postpone their transmissions or keep quiet, which is the best they can do. When it comes to the network layer, this would imply that there is no redundancy left to deal with packet loss since all 2f+1 honest nodes would have to reach the final state in order for the transaction to finish, and in this instance, packet loss would be catastrophic. When the number of redundant nodes is increased to more than 3f+1 nodes, we get the same regimes as those described previously
Even if 5f+1 nodes are utilized, we achieve comparable success percentages in the worst scenario, since such a system would need a 3f+1 quorum and leave 2f overall redundancy in the system, i.e., f Byzantine nodes and f honest nodes to whom the message does not need to be sent. It should be noted that this is only true if the adversary does not have access to the channels between honest nodes, which was the underlying assumption we began with. Alternatively, if the packet loss or the frequency of node failures is too high for UDP to be used, the implementation may always fall back to TCP and therefore imitate reliable channels over unreliable channels. The safety characteristic of the system is never jeopardized, and only performance is increased in too optimistic circumstances, according to the system’s design.
5. Conclusion
The distributed ledger technology assists us in the establishment of a peer-to-peer-based secure network comprised of a conglomeration of the owners of parking lots, garages, users searching for parking, and free parking spots. The network allows those who own parking spots to make use of their spaces while also earning money by registering their places in the network. The directed acyclic graph assures that consumers get the most cost-effective service possible by assigning unique consensus timestamps to each and every service request. We spoke about how a dynamic pricing model, which generates unique charges in response to each individual parking request, would be advantageous to both parking lot owners and those looking for parking spaces. By using this strategy, consumers may save both time and money while also assisting parking lot owners in getting the most out of their available parking spots. The results of the simulation demonstrate that the suggested model is effective in eliminating jams, reducing users’ search times, and ensuring maximum usage of available resources, among other things. The increase in resource utilization would go a long way toward enhancing the incentive of small parking lot owners and private parking space owners to provide parking space for customers. Users will soon have more parking alternatives, and the likelihood of a traffic bottleneck will be reduced as a result of this. The parking space allocation use case shown in this chapter is just a representative example of what may be done. On the one hand, we can observe that the hash graph maintains the order or transactions in terms of entrance and exit, but the general blockchain does not retain the order or transactions.
Data Availability
The data that support the findings of this study are available on request from the corresponding author.
Conflicts of Interest
All authors declared that they do not have any conflict of interest.
Acknowledgments
The authors extend their gratitude to the Deanship of Scientific Research at King Khalid University for funding this work through the research groups program under grant number R. G. P. 1/85/42.