Introduction to the Development of Intelligent Parking Systems
Automated parking system
An early form of intelligent parking assist systems was the Automated Parking System (APS), developed in response to the rapid increase in urban vehicle numbers and the need for efficient parking solutions APS aims to maximize parking capacity by efficiently utilizing available space and attracting customers, effectively reducing the area or volume needed to park cars Modern automated parking systems not only optimize space management but also assist operators in maximizing parking efficiency, while helping drivers find vacant slots quickly, saving time and reducing costs Additionally, these systems contribute to lowering vehicle emissions by minimizing the time spent searching for parking spots.
Since the motor vehicle was born, an inevitable implication which is pulled by the parking system also born The early parking systems were mechanical systems with the
The development of automated parking systems has evolved significantly over time, beginning with the introduction of an automated parking system in Paris, France, in 1905 at Garage Rue de Ponthieu, marking a milestone in parking technology As science and technology advanced, the 1990s saw a surge in parking infrastructure, with Japan constructing nearly 40,000 parking spaces annually using the paternoster Automated Parking System (APS) By this period, approximately 1.6 million parking spaces were being built, highlighting the widespread adoption and importance of automated parking solutions in modern urban development.
In 2012, rapid urbanization and the growing number of used cars created a pressing demand for more efficient parking solutions To better meet user needs, traditional parking lots increasingly transitioned to Automated Parking Systems (APS), offering enhanced convenience and space utilization.
Automated parking systems (APS) utilize advanced technology similar to mechanical handling and document retrieval systems, enabling efficient vehicle management In these systems, drivers simply leave their car in an entrance module, and a robotic trolley transports it to an available parking spot, saving time and reducing the stress of finding parking APS offers key features such as minimizing vacant slots, assisting drivers in locating parking spaces, and supporting automated payment and charging processes The benefits of automated car parks include improved space utilization, enhanced convenience, and increased operational efficiency.
• Utilize the parking spaces in the parking lot by having plans to use vacant spaces
• Increase the capacity of the parking lot used by reducing the width and height of each parking space
• More secure because the system prohibits public access to the parking lot
• Reduce driving time and cost because the search and move to the parking spot inside car park are eliminated
• The visual impact and handicap access are improved
• Shorter waiting time compare with traditional parking system
Automated parking systems (APS) face challenges such as high installation and maintenance costs, limited flexibility in organization, and low user popularity due to technical difficulties in building fully functional mechanical systems Their placement in crowded areas often leads to congestion at entry points, causing excessive wait times and reduced service speeds To address these issues, intelligent parking assist systems (IPAS) have been developed, offering greater flexibility by utilizing new technologies that enable installation in various locations, including traditional and automated car parks IPAS leverage IoT technologies to enhance user interaction, allowing drivers to easily find parking spaces through smart devices, thereby increasing customer engagement and accessibility These systems improve marketing opportunities and support the creation of widespread parking network plans for greater efficiency This study emphasizes the characteristics of current IPAS and proposes a new network model to promote future development of intelligent parking solutions.
Intelligent parking assist system
Historically, many parking systems were developed without considering the user's perspective, making it difficult for drivers to find their ideal parking spot The introduction of intelligent parking assist systems has been a significant advancement, enhancing driver convenience and streamlining the parking experience These innovative solutions now help drivers locate the most suitable parking spaces quickly and efficiently, reflecting a user-centric approach in modern parking technology.
Finding a good parking spot quickly is especially important when you're in a rush, such as before an important work event Traditional parking systems often make it difficult to locate suitable parking spaces efficiently, leaving drivers to rely on luck In contrast, intelligent parking assist systems (IPAS) leverage advanced technologies and wireless communication to guide drivers to the best parking spots with just a few simple operations via a smartphone or screen IPAS automatically helps drivers find suitable parking spaces based on their profiles, minimizing time and cost, and enhancing overall parking convenience and efficiency.
The IPAS (Intelligent Parking Assistance System) was first introduced in 2003, enabling automated parking without driver intervention through front and rear cameras and fixed sensors Early versions faced challenges in accurately detecting objects like pedestrians and small obstacles, and would issue continuous warnings in tight spaces In 2005, system upgrades added recognition of parking stripes, while 2006 saw the integration of parking sensors to enhance functionality However, initial systems lacked advanced features like Parking Information System (PIS) and Parking Guidance System (PGS) Over recent years, leading automakers such as BMW, Mercedes, Audi, and Toyota have developed advanced IPAS models incorporating automation, real-time notifications, and driver assistance to simplify parking and improve safety The primary goal of modern IPAS systems is to help drivers efficiently find suitable parking spaces, reducing time and enhancing convenience.
An IPAS system, featuring components like Smart Navigation and Intelligent Parking Assistant, offers an affordable solution for efficient parking management The Smart Navigation utilizes an online traffic simulator to deliver real-time traffic updates and optimized route maps to reserved parking lots, enabling drivers to reach their destination quickly The Intelligent Parking Assist continuously provides updates on parking availability, helping drivers avoid full or busy lots and saving valuable time Additionally, this system gathers driving behavior data—such as aggressiveness, reaction times, and following distances—to improve travel time estimates and enhance overall navigation accuracy.
Figure 1 2 The basic configuration of an IPAS system
Classification of intelligent parking assist system
Since the introduction of intelligent parking systems, various types have been developed, including Parking Guidance and Information Systems (PGI), which help drivers find available parking spots efficiently, and Parking Reservations Systems (PRS), allowing users to reserve parking spaces in advance Other notable implementations include comprehensive Intelligent Parking solutions that integrate multiple components to optimize parking management, enhance user experience, and reduce congestion in urban areas These systems leverage advanced technologies such as sensors, IoT, and real-time data to improve parking efficiency and convenience.
Assist System (IPAS) is a comprehensive hybrid system built on Internet of Things (IoT) technologies to address parking challenges efficiently Many Parking Guidance and Information (PGI) systems have been developed to help drivers find available parking spaces quickly, reducing search and travel times, and alleviating road congestion, fuel consumption, and CO2 emissions These systems typically communicate parking availability via instant messages through vehicle network protocols However, PGI systems lack statistical data and forecasting capabilities for vehicle flow, often leading to congestion and increased waiting times—phenomena known as “many cars chasing a parking lot.” To mitigate this, Parking Reservation Systems (PRS) were introduced, allowing drivers to book guaranteed parking slots via SMS, which reduces congestion, improves traffic forecasting, and enhances revenue With advancements in electronic, IoT, and cloud computing technologies, the innovative IPAS offers a full spectrum of parking services, integrating these technologies for smarter parking management The detailed architecture of IPAS will be elaborated in Chapter 2 of the thesis.
Shortfalls of current Parking Systems
The current parking systems mentioned in above are encountering some of shortfalls for each system, these problems come from the structure of them
The shortfalls of PGI systems are:
• PGI systems provide users with parking information comparable to older parking lots, but this information is limited and not a complete solution
• PGI systems assist users in finding vacant parking spaces, which helps to minimize congestion and drive time However, this information is localized,
7 leading to the phenomenon of many cars chasing the same parking lot and may lead to congestion or increase the average waiting time for parking
Operation and maintenance costs of PGI systems are relatively high due to the limitations of their sensor technology, especially in indoor guidance applications These sensors are highly sensitive to temperature and weather fluctuations, which can affect their performance Additionally, the sensors rely on battery power, requiring regular recharging to ensure continuous operation.
PGI systems rely on limited technologies, which prevent them from delivering optimal solutions for parking management and services Consequently, parking information is restricted to the availability status of individual parking lots, limiting overall efficiency and user convenience.
The shortfalls of PRS systems are:
The primary limitations of the current reservation system include its restriction to limited reservation periods, which constrains flexibility for users Additionally, the system utilized a fixed pricing model without estimating potential revenue, and only considered a single destination option Long-term parking reservations, such as week-long bookings for events, could significantly enhance user appreciation However, city councils and private parking operators may hesitate to invest in such systems unless they demonstrate a clear improvement in parking revenue.
Agent-based personalized rating systems (PRS) require significant advancements to enhance their effectiveness Integrating machine learning techniques into car park and driver agents can enable more flexible and dynamic bid adjustments, improving system responsiveness Additionally, equipping driver agents with advanced reasoning capabilities to assess driver behavior can lead to more accurate and fair ratings However, current agent-based PRS frameworks remain immature and are not yet ready for widespread real-world deployment.
Pricing-based reservation systems can create fairness challenges, as free or cheaper parking spaces tend to be reserved predominantly by wealthier individuals, leading to unequal access Additionally, lower-cost parking lots often experience increased congestion due to higher demand from drivers seeking affordable options, which can exacerbate traffic issues and reduce overall efficiency.
• Mobile/Web Parking Information and Reservation Systems reviewed earlier indeed reduce the overall parking problems However, they should not be utilized
8 on their own Instead, a smart parking reservation management model should be integrated to them, such that a general objective (e.g., maximize revenue or minimize cost) is achieved
Recent advancements have introduced IPAS hybrid systems with fully integrated PGI and PRS functions, offering a flexible and user-friendly architecture These systems enable operators to seamlessly integrate local parking facilities into extensive parking networks by leveraging existing infrastructure and supplementary devices This integration facilitates the creation of comprehensive global parking information, significantly supporting intelligent transportation systems and enhancing urban parking resource management.
Architecture for Intelligent Parking Assist Systems based on
Internet of Things technologies
The Internet of Things (IoT) is a revolutionary inter-networking technology that connects billions of smart electronic devices, including smartphones, sensors, vehicles, and building systems, through internet connections Before IoT's development, these devices were passive and unable to share information with each other With the advent and application of IoT technology, these devices became "smart devices," enabling seamless communication and data exchange among them.
The term "Internet of Things" (IoT) was first introduced by Kevin Ashton in 1999, later shaping the definition provided by the IoT-GSI in 2013, which describes IoT as a global infrastructure that interconnects physical and virtual objects to enable advanced services through interoperable communication technologies An IoT "thing" can be any physical or virtual object capable of identification and integration into communication networks Experts project that by 2020, approximately 34 billion devices will be connected to the Internet, with 24 billion being IoT devices, facilitating innovative applications such as smart parking systems that monitor parking lot availability using connected electronic devices.
Intelligent IoT-based parking systems enhance urban mobility by enabling users to pre-reserve parking slots remotely via secure smartphone applications like E-parking, which utilizes low-cost IR sensors and Raspberry Pi 3b for real-time data collection, effectively reducing traffic congestion and providing cost-effective, authenticated parking solutions These systems, such as those developed in studies [39, 45], use interconnected hardware like Raspberry Pi, distance sensors, and Pi Cameras to gather data on vehicle count, license plates, and parking availability, transmitting this information to cloud servers for instant user access through smart devices, thus saving time and lowering CO2 emissions Additionally, parking guidance software leveraging MQTT protocol ensures real-time, efficient sharing of parking lot status updates across numerous clients, making the system scalable and energy-efficient while enhancing urban environmental sustainability.
Many types of IoT-based intelligent parking systems have been implemented, enhancing parking efficiency and user convenience In the next section, we will analyze and introduce the fundamental components that make up these IoT-driven parking solutions, highlighting their role in transforming urban parking management.
Structure of an IPAS System based on IoT technologies
Figure 2 1 The basic structure of an IPAS system based on Internet of Things technologies
An intelligent parking assist system based on Internet of Things technologies typically consists of 3 main components: Local Parking Units, System Database, Software Client [22-23]
A local IoT network in parking lots integrates wireless sensor networks and control units to provide real-time parking space information This network enables efficient management by monitoring available spots and displaying data to drivers Installed within each parking facility, the system's main components include sensors that detect vehicle presence and control units that process and display parking data Implementing such IoT solutions improves parking management, reduces congestion, and enhances user experience through seamless information sharing.
The control unit, such as an Arduino module, serves as the central control module connected to sensors to monitor parking lot occupancy Various solutions are available, including RFID card readers that track cars entering and leaving the system, as demonstrated in reference [23], and infrared sensors installed at each parking space to detect availability These technologies enable accurate car counting and real-time parking space status updates.
The control unit for parking spaces gathers real-time data on availability, service rates, and other relevant information to monitor the parking lot effectively It transmits this data to a central database server via an internet connection, enabling demand prediction and improving driver guidance Acting as a gateway between the local IoT sensor network and the database, the control unit also performs local parking lot controls Currently, hardware platforms like Arduino and Raspberry Pi are commonly used as control units in IPAS systems to ensure efficient parking management.
RFID tag, ID card Enteri ng
4 St atus of checking RFID t ag
RFID tag, ID card Departure
Figure 2 2 The role of the control unit in an IPAS system
IoT sensors play a crucial role in monitoring car park availability by detecting the percentage of free parking spaces These sensors, which may include RFID tags, provide real-time data to improve parking management efficiency Implementing IoT technology ensures accurate, up-to-date information on available parking spots, reducing search time and enhancing user experience.
ID card, IR sensor, Acoustic sensor, Magnetic sensor, Ultrasonic sensor [1, 23, 25].
(a) The priciple of ultrasonic sensor
(b) Two ultrasonic sensor for detecting a passing car
Figure 2 3 The use of ultrasonic sensor to detect the presence of the car at the parking space
The screen serves as the front-end interface of the parking management system, providing essential information such as the capacity of the local car park, the current percentage of available spaces, and the status of RFID tag checks It also features an indoor mini map for navigation, giving drivers a comprehensive overview of the parking lot to enhance their parking experience.
This server is designed to store online parking lot data transmitted by the control unit, including real-time information on parking availability and current In/Out traffic By providing accessible and up-to-date parking data, it enables drivers to easily search for and find parking information without accessing local parking facilities directly When the parking lot data volume is minimal, a traditional centralized database server can efficiently support this system.
14 server [24] For large systems with massive data, for example, a network of parking lots as introduced in Section 2.4, there are solutions: based on Cloud computing server [23-
Recent developments in large-scale parking system architectures highlight the shift from traditional cloud computing to fog computing technology Cloud-based systems often face challenges such as high latency due to processing large data volumes, bandwidth limitations caused by transmitting every data bit over network channels, and slow response times with scalability issues stemming from reliance on remote servers Fog computing addresses these problems by operating at the network edge, reducing processing time, minimizing bandwidth usage through local data aggregation at access points, and delivering sub-second response times Additionally, fog computing enhances system scalability, reliability, and fault tolerance, making it an effective solution for large-scale parking management systems.
Figure 2 4 Propose to use Fog Computing in building intelligent parking assist system
This article compares Cloud Computing and Fog Computing for IPAS systems based on IoT technologies, highlighting their respective advantages Due to the real-time data requirements and the need for extensive storage and computing power in large IPAS applications, Fog Computing is proposed as a more suitable solution A study cited in [32] demonstrates a simple Fog Computing-based software architecture for NB-IoT networks, successfully supporting IoT applications like intelligent parking systems, illustrating the practical application of Fog Computing in IoT-driven IPAS solutions.
Table 2 1 Comparison between Cloud Computing and Fog Computing for IPAS system
Parameters Cloud Computing Fog Computing
Delay of Service Normal Very low
Location of Service Within the Internet
At the edge of the local network (locate at each local IoT parking lot)
Number of hops between client and server Multiple hops One hope
Security Undefined Can be defined
This application is designed for smartphone platforms such as Android and iOS, enabling users to access a comprehensive range of services needed by drivers [17, 20, 24-25] Most smart devices operate on either Android or iOS, so IPAS applications are specifically developed for these platforms For Android-based IPAS applications, developers utilize Android Studio, while Xcode is used for creating applications on the iOS platform.
Figure 2 5 The interface of an IPAS system running on iOS platform
IoT Hardware for IPAS Systems
Parking space monitoring sensors are categorized into ‘on-street’ sensors placed directly on the parking lot surface and ‘off-street’ sensors positioned at a height above ground On-street sensors, such as pneumatic tubes, loop detectors, magnetic sensors, piezoelectric sensors, and RFID, can be either embedded into the pavement or taped onto the surface for accurate parking detection Off-street sensors include ultrasonic sensors mounted at elevated positions, providing effective monitoring solutions from above These sensors play a crucial role in modern parking management systems by enabling precise detection and efficient space utilization.
Pneumatic road tubes are installed perpendicular to traffic flow and are widely used for short-term traffic counting A pair of tubes can span multiple lanes, allowing for effective vehicle detection The data logger records which tube is crossed first to determine vehicle direction; however, this method has limitations, as simultaneous vehicle crossings can lead to inaccurate direction identification.
A loop detector is an inductive loop embedded in or beneath the road surface, utilizing the principle that a magnetic field near an electrical conductor induces an electrical current When a vehicle passes over the loop, it interrupts the magnetic field, causing a reduction in inductance This change in inductance allows the loop detector to identify the presence of vehicles, making it an essential component in traffic management and automated toll systems.
17 then discovered by a control unit which produces a signal to a processing computer module either in wired or wireless form that by its subscribes the inductance variations as vehicle counts
Magnetic sensors detect changes in the Earth's magnetic field caused by passing vehicles, enabling traffic monitoring These sensors are typically installed either buried under the road surface or housed in roadside boxes However, a key limitation of magnetic sensors is their difficulty in distinguishing between closely spaced vehicles, especially when two cars are traveling in close proximity.
An acoustic sensor detects vehicles by analyzing sound energy generated through the interaction between tires and the roadway It uses microphones to monitor changes in noise levels, converting these variations into signals that indicate the presence of a vehicle This reliable detection mechanism leverages sound energy analysis to accurately identify vehicle movements on the road.
A piezoelectric sensor detects vehicle movement by converting mechanical energy into electrical signals, making it ideal for traffic monitoring Mounted within a groove in the road surface, the sensor generates a voltage when a vehicle drives over it, with the signal's amplitude proportional to the degree of deformation As the vehicle passes, the voltage reverses when it moves off, providing accurate vehicle detection and counting This reliable, contactless sensing method enhances traffic management systems and roadway analytics.
RFID technology is one of the most widely used devices in modern parking systems, comprising an RFID reader and RFID tags with unique identifiers for vehicle identification Similar to barcodes or magnetic strips on credit cards, RFID tags must be scanned to retrieve their information, enabling efficient vehicle authentication In IPAS systems, RFID tags are assigned to vehicles; when a vehicle enters the parking lot, the RFID reader scans the tag, verifies its registration, and grants access if authenticated Successful identification updates parking data, including vehicle count and parking duration, ensuring seamless and secure parking management.
Ultrasonic sensors measure distance by emitting ultrasonic waves and receiving the reflected signals from a target, such as a vehicle They determine the distance by calculating the time elapsed between wave emission and reception, providing reliable and accurate measurements Due to their simplicity and precision, ultrasonic sensors are widely used in car parking systems to detect parking space occupancy, enhancing automation and parking management.
Infrared sensors, including active and passive types, are utilized for vehicle detection Active IR sensors transmit and measure the reflection levels of infrared light, capturing snapshots of the reflected signals from the roadway surface; any changes in these reflections indicate the presence of a vehicle Passive infrared sensors detect vehicles by measuring the infrared radiation emitted from the detection zone, with an increase in radiated energy signaling that a vehicle has passed through.
2.3.2 Arduino, Raspberry Pi Control Units
Arduino is a versatile microcontroller with input and output capabilities, making it ideal for prototyping and small IoT projects It is commonly used as a control unit to collect sensor data and monitor parking space occupancy, often integrated with Arduino Ethernet for internet connectivity, enabling real-time remote monitoring For example, the Arduino Uno platform has been used to display parking slot availability via LED lights—green for free, yellow for reserved, and red for occupied slots—in smart parking systems Additionally, Raspberry Pi is frequently employed as a control platform in IoT parking solutions, providing robust processing power for more complex applications.
The Raspberry Pi, depicted in Figure 2.6, is a fully functional mini-computer known for its compact size and low power consumption, making it an ideal platform for IoT systems It is highly versatile and powerful, often employed in implementing IPAS systems, as demonstrated in studies [37-40] While simpler IPAS systems typically utilize the Arduino platform due to its ease of use, the Raspberry Pi is preferred for more complex tasks that demand greater processing capabilities and advanced functionality.
Arduino Module Raspberry Pi module
Figure 2 6 Andruino/Raspberry Pi module used as control unit in Intelligent Parking
Protocol stack for IPAS systems
Based on an analysis of current IoT technologies, we propose a four-layer protocol stack for IPAS systems, as depicted in Figure 2.7 The protocol stack is organized into the Perception, Networking, Middleware, and Application layers, each serving a specific function The Perception layer collects and senses data from IoT devices, enabling real-time data acquisition The Networking layer ensures reliable communication and data transfer between devices and systems The Middleware layer facilitates data processing, management, and integration across different platforms Finally, the Application layer delivers user-centric services and interfaces, supporting the deployment of IoT solutions in various industries This layered architecture enhances the efficiency, scalability, and interoperability of IPAS systems within the IoT ecosystem.
The Perception layer comprises hardware control units such as Arduino and Raspberry Pi, along with sensors like IR, ultrasonic, and acoustic sensors to detect parking space occupancy It also includes RFID tags/readers or ID cards for vehicle authentication and counting vehicles entering and exiting the parking lot Its primary goal is to monitor real-time parking information, which is transmitted to higher layers to deliver efficient parking services to drivers.
The networking layer in IoT systems utilizes key protocols such as HTTP, MQTT, and CoAP to support diverse device communications within a heterogeneous IPAS network These protocols are essential for enabling machine-to-machine (M2M) and real-time IoT applications HTTP is primarily designed for traditional internet devices with ample power sources, providing a reliable method for data transfer, while MQTT and CoAP are optimized for constrained devices, ensuring efficient communication in resource-limited environments.
HTTP is commonly used in IPAS systems for connections from user devices and control units to the server system, particularly when sensor nodes have high processing power and can send data directly to the internet For constrained network nodes such as sensors in IPAS systems, protocols like CoAP and MQTT are more suitable due to their efficiency in low-power environments All three protocols—HTTP, CoAP, and MQTT—support IPv6 addressing, accommodating the large number of sensors typical in IoT networks A comparative analysis of these protocols in IPAS system construction is detailed in Table 2.2.
Table 2 2 Comparison of protocols used in IPAS system
Feature RESTful HTTP MQTT CoAP
Transport protocol TCP TCP UDP
Network layer IPv4/IPv6 6LoWPAN/802.15.4 6LoWPAN/802.15.4
Architecture model Client-Server Publish-Subscribe Client-Server/
Synchronous communication Require Require Not require
Application Internet devices Internet devices
Multicast support No Yes Yes
The middleware layer is essential connectivity software that provides a set of enabling services, allowing multiple processes on one or more machines to interact seamlessly across a network It acts as a crucial intermediary in client/server architectures, ensuring reliable communication and integration between distributed systems By facilitating efficient data exchange and process coordination, the middleware layer enhances system scalability and performance in networked environments.
21 applications such as IPAS applications and providing for communication across heterogeneous platforms
• Application layer: this layer provides interfaces for users to access IPAS system resources through a web browser or through client software
Figure 2 7 Stack of protocols for Intelligent Parking Assist System
Propose the IPAS network based on IoT technologies
Many current proposals for building intelligent car parking systems lack comprehensive functionality, often failing to assist users in finding available parking slots across multiple locations Most existing solutions either offer limited assistance within a single parking lot or provide only partial guidance, indicating that there is significant potential for developing more integrated and effective parking management systems.
22 parking network In order to solve these problems, in this work, we propose a novel parking network architecture that links the parking lots together and optimize parking resources
In our study, we proposed utilizing the Car Park Network (CPN) architecture as the backbone infrastructure, featuring a combination of wireless and wired links The CPN infrastructure includes routers that act as the central connection point for both connected clients and sensor networks, enabling seamless communication This self-configuring and self-healing network can connect to the Internet via gateway functions, supporting infrastructure meshing to integrate the CPN with existing Wireless Sensor Networks (WSNs) Additionally, conventional clients equipped with compatible radio technologies can directly communicate with CPN routers, ensuring versatile connectivity within the parking network ecosystem.
Figure 2 8 Infrastructure/Backbone of the CPN architecture
Each car park is modeled as a node within a comprehensive parking network The real-world deployment of this network is illustrated in Figure 2.9, which visually depicts each parking facility labeled accordingly This approach enables efficient analysis and management of parking resources across the entire system.
Figure 2 9 Car park network deployment for car parking system
• P1 is car park number 1, N1 is total parking spaces in P1
• P2 is car park number 2, N2 is total parking spaces in P2
• Pn is car park number n, Nn is total parking spaces in Pn
The total system capacity is represented by N, which is the sum of individual capacities (N = N₁ + N₂ + N₃ + … + Nₙ) The real distance between two nodes in the network is denoted as D, while Dij specifically indicates the distance between node Pi and Pj Building upon the work in [23], M symbolizes the actual parking fee at each car park, with Mj representing the hourly parking fee at car park Pj Figure 2.10 provides a visual illustration of our network design.
Each node is equipped with a neighbour table to monitor the current network status and a queue with a predefined capacity to manage vehicle flow, preventing overloads The neighbour table stores information about directly connected nodes, while the queue controls the number of vehicles forwarded to the node, ensuring it operates within its capacity When a node joins or leaves the network, it broadcasts a message containing its available resources to neighboring nodes, which update their neighbour tables accordingly In this network, node capacities are defined as N1=100, N2=120, N3=200, N4=100, N5=120, N6=120, and N7=100 spaces, with distances between nodes such as D12=1.2 km, D13=1.6 km, D23=2.0 km, D27=1 km, D34=1.5 km, D37=1.8 km, D45=1.2 km, D56=0.8 km, and D67=1.2 km Parking at car park 1 costs $1 per hour.
This study evaluates parking fees across different car parks, with rates ranging from $0.5/hour for Car Park 6 to $2.00/hour for Car Park 4, as depicted in Figure 2.11 using neighbor tables The parking lot capacities are defined, with N1 hosting 20 free spaces, N2 and N3 each with 60, N4 with 70, N5 with 60, N6 with 30, and N7 with 60 To enhance the efficiency of locating available parking, each node's neighbor table includes current data on free parking spaces in adjacent nodes The proposed approach calculates the cost of selecting a parking lot based on the total number of free spaces available in each node, optimizing resource allocation.
2.5.2 Construct the neighbor table of nodes
We assume a network with seven nodes as in Fig 2.10 and calculate the value of cost function F (,,) (see Section 3.1) with α = 0.2, β = 0.6, γ = 0.2, D = 2 km, T
The neighbor table for each node, based on the F (α, β, γ) function, is detailed in Fig 2.12, demonstrating its alignment with Eq (3.1) This routing table is essential for selecting the next node to forward the user to when a parking lot is full, ensuring efficient navigation within the network.
Figure 2 12 Neighbour tables sorted by descending values of F ( , , )
Figure 2.13 demonstrates how users search for parking lots by evaluating the cost function, which considers factors like distance from their location, parking fees, and available vacant spaces The system identifies Car Park 2 as the optimal choice based on these criteria Detailed insights into the parking cost function estimation will be provided in the next chapter, highlighting its importance in optimizing parking solutions.
(3) ResMessage: Best Car park 2, parking slot 2, Distance 2km
ReqMessage: Which is best car park?
Car park F(α,β,γ ) Dis Capacity %full P1
Figure 2 13 A searching for a parking space based on the cost function
Mathematical Models for IPAS Systems
Linked-Cost function estimation
When searching for a parking space, users consider several key factors such as the shortest distance from their current location, the lowest parking fees, and the availability of free spaces, which directly impacts waiting times Often, drivers evaluate a combination of these criteria—distance, cost, and space availability—to make the most efficient choice To address these preferences, we utilize an extended function, F(α, β, γ), which integrates these factors and provides a comprehensive decision-making tool for optimal parking spot selection.
Equation (1) in study [23] to calculate the cost between the nodes in the network
F is a function that depends on the distance between two nodes, the number of free parking spaces at the destination node and the parking fee at the destination node
F is considered to be a weighted link between two nodes in the parking network
When two nodes are not directly connected, the function F(α, β, γ) is assigned an infinite value If a vehicle arrives at a node that is full, it is forwarded to the neighboring node with the smallest F(α, β, γ) value in the neighbor table, ensuring efficient routing as described in Section 2.5.2.
We calculate the cost function F ( , , ) from node P i to node P j :
( ) up j up j up ij ij ij M m T t D
The function F = α, β, γ = α × + β × + γ × encapsulates key factors influencing node connectivity Specifically, α depends on the path length between nodes, β pertains to the number of available free slots at the destination, and γ relates to the parking fee per hour This function is inversely proportional to the distance between two nodes, indicating that shorter paths increase connectivity Additionally, F is directly proportional to the total number of free slots at the destination, highlighting the importance of available capacity in the network performance Understanding these relationships helps optimize node interactions based on distance, available slots, and parking fees.
Adjusting parameters α, β, and γ allows for optimizing network performance based on the importance of distance, free parking slots, and parking costs When α is set to 0, the model ignores distance and focuses on free parking spaces and costs; setting β to 0 means only distance and parking fees are considered, while γ set to 0 emphasizes distance and total free parking spaces without factoring in parking charges These experiment-derived parameters, ranging from 0 to 1, enable tailored adjustments for efficient parking network management and improved user experience.
In equation (3.1), we determine the cost function by considering the distance between two nodes and the percentage of available parking spaces at each location The calculation incorporates the maximum possible distance between nodes and the maximum parking capacity of each parking lot, ensuring an accurate assessment of parking efficiency and accessibility.
In the model, the distance between nodes \( P_i \) and \( P_j \) is denoted as \( d_{ij} \), with \( D_{up} \) representing the maximum allowable distance, a global parameter The variable \( t_j \) indicates the number of parking spaces occupied at node \( P_j \), while \( T_{up} \) sets the overall parking capacity limit for the entire network Parking fees at each node \( P_j \) are represented by \( m_j \), with \( M_{up} \) serving as the global fee cap for the entire parking system.
We also used a function Fcost to evaluate the fee that user will pay in term of parking fee to have serviced
( , , ) ( ) ( ) ( ) cos cos F Cost Cost Cost
Cost(α), Cost(β), and cos(γ) represent the respective cost functions related to the variables α, β, and γ These values are mapped to user fees based on empirical data, as shown in Table 3.1, where lower values of α correspond to lower user fees, aligning with user preferences for shortest travel distances to parking Similarly, β has smaller values due to the greater availability of parking spaces, which typically result in lower charges Conversely, γ has the smallest value, indicating that higher probabilities of γ reflect increased user interest in parking facilities with higher fee rates.
Table 3 1 Table of cost function
Cost function Cost(α) Cost(β) Cost(γ)
From the cost function, we can deduce the possibilities for users to search for suitable parking places:
Users prefer to find the nearest car park for convenience, which is determined by calculating the shortest distance between the user and surrounding parking options This distance is computed using the GPS coordinates of the user, denoted as \( u_i (x_i, y_i) \), and those of the car parks, \( p_j (x_j, y_j) \), where \( x \) represents latitude and \( y \) represents longitude By applying this method, the system effectively identifies the closest car park based on geographic location.
The total number of suitable car parks in the area is denoted as N, with j ranging from 1 to N By setting β = 0 and γ = 0 in Equation (3.1), we can derive a specific result that corresponds to the minimum value of the function F(α, β, γ) within the neighbor table This approach simplifies the analysis by focusing on the optimal parking options based on the given parameters.
• Least Cost Users want the park with the lowest cost per hour at their destination
The system retrieves the appropriate result by performing a database search and calculations Specifically, setting α = 0 and β = 0 allows us to obtain a result from Eq (3.1) This result represents the minimum value of the function F(α, β, γ) within the neighbor table, as discussed in reference [24].
An optimal car park, offering a balance of suitable distance, price, and a high likelihood of available spaces, enhances user convenience by minimizing wait times The system efficiently distributes parking demand across the network, preventing congestion at popular spots and reducing the parking bottleneck phenomenon This approach not only improves user experience but also increases profits through the effective utilization of available parking spaces To achieve optimal results, we selected specific parameter values (α, β, γ) based on simulations and implemented outcomes, ensuring the minimum of the function F(α, β, γ) in the neighbor table as detailed in [24].
Total Parking Cost Estimation
We build the mathematical models of our proposed system based on the results in
We develop an effective parking planning strategy by defining key parameters such as S, representing the total number of available parking lots, and W, a set containing the costs (wij) between each vehicle and parking facility The cost wij is determined based on the GPS distance from vehicle pi (where pi belongs to the set of vehicles P) to parking lot Sj (in set S), as well as the current availability of free spaces in Sj This approach allows us to optimize parking allocation by accurately assessing transportation distances and parking lot occupancy.
M and N be the size of P and S, respectively Therefore, the size of W is MxN
Assuming that vehicles are jobs and parking places are servers, Wij is the cost for server Sj doing job Pi We save the solution in X where 𝑥ij ∈𝑋 That is,
S at park not will P if , 0
S at park will P if , 1 x ij (3.4)
Let C be the total cost for all vehicles in P going to the parking places assigned to them by the IPAS That is,
In here, we use F (,,)as the cost so that we have a new total cost:
To reduce user costs, we select the minimum value of F(α, β, γ) from equation (3’) Our goal is to minimize the total cost C while ensuring that each vehicle is allocated exactly one parking resource, and each parking space is assigned to only one vehicle.
1 indicates that any user in the queue may be assigned at most one car park but may also fail to get an assignment On the other hand,
1 still guarantees that each user in the queue maintains a car park assignment
Our proposed system intelligently manages parking availability by suggesting alternative parking lots when a vehicle arrives at a full car park If no free spaces are found, the system forwards the vehicle to another nearby parking facility, enhancing parking efficiency and reducing congestion Let h represent the number of vehicles that are forwarded to different parking lots Each parking lot, denoted as Sj, is assigned a capacity for kj vehicles, where kj corresponds to the total number of available free slots in that lot The total capacity across all parking lots is represented by the sum of all kj values, ensuring optimal utilization of parking resources.
Our algorithm has a time complexity of O(n*k), optimizing the efficiency of parking resource allocation We aim to minimize the time required for each vehicle to locate a free parking spot, enhancing overall performance By implementing an effective mathematical model, we have reduced total costs and achieved a more efficient distribution of users across the network resources.
Parking Queue Models
Based on our previous works in [23], we modelled the system into a service queue
It includes all users entering at each car park The entering process at each node is considered to be a FIFO queue and Markov process
The modeling of the service queue is represented by the M/M/1/K/FIFO system, where the first “M” indicates that arrivals follow a Poisson (Markovian) distribution, and the second “M” signifies that service times are exponentially distributed The system consists of a single server (1) and a finite capacity of K lots Key parameters include the inter-arrival time (μ_A) and the service time (μ_S), with the assumption that μ_S is significantly larger than μ_A to prevent queue instability The average waiting time in the queue for a long-term perspective in an M/M/1 system is a critical performance measure.
The average waiting time in the queue (expected from a long run) in the M/M/k queue is:
The average waiting time of the system is:
= = 1 where N is the total number of parking spaces
Smart Services System Over Parking Networks
Introduction
Recent proposals for intelligent car parking systems aim to improve user assistance in finding free parking spaces, but many lack full functionality across entire parking networks In [49], a novel daytime and nighttime vacant parking space detection system was developed, enhancing detection robustness through an improved Bayesian hierarchical framework (BHF) that models parking lots as a 3D domain with planar surfaces This method employs plane-based classification and a modified BHF to accurately detect available parking spots, regardless of lighting conditions However, the study focused on a single parking lot using camera-based detection and did not address building comprehensive smart parking networks with integrated services.
This article discusses the evolution of parking management systems, highlighting a 36-car parking framework that monitored vehicle entries and exits using sensors and magnetic boards to calculate available parking spaces, which were then displayed to users prior to entry However, this system only informed users about free spots without integrating parking plans or optimizations In contrast, a more advanced smart parking system was proposed for urban environments, which reserves and assigns optimal parking spaces based on a cost calculation that considers proximity to destinations and parking fees, enhancing efficiency over traditional guidance systems Although this system allows drivers to reserve both on-street and off-street parking, it lacks realistic implementation, comprehensive services, and guidance features to help drivers find the best available parking quickly en route to their reserved spot Therefore, the proposed framework in this chapter aims to provide a complete set of services necessary for future intelligent parking assist systems, improving overall parking management and user experience.
System Framework
Figure 4.1 describes our system framework in the view of parking services In figure
2, we can see that the proposed system is separated become three layers: Service, Core, and Management layer
Figure 4 1 System framework for IPAS services
The service layer serves as the primary interface for user interaction, providing essential functionalities such as guidance, booking, tracking, warnings, and car locking services These features are designed to support drivers in quickly reaching their desired parking spots, ensuring a seamless user experience To access these services, users must log in via an Android smartphone, with data exchanged securely through 3G, 4G, or Wi-Fi connections This comprehensive service layer is crucial for delivering reliable and efficient parking management solutions.
The core layer of the system comprises essential components such as the GPS information center, Calculation center, and Decision and Analysis center, which work together to execute user commands and service requirements These centers rely on data from the Parking Information block, Parking Traffic State updating block, Control Parking Arduino Unit, and RFID Reader Card ID Information to deliver accurate and efficient parking management solutions.
Core components retrieve data from the database then calculate, analyze, and return the results to the requested services
The GPS Information Center provides real-time location data based on the user's GPS address The Calculation Center determines the total cost for the user by utilizing data from the GPS Information Center and identifying the nearest parking facility The Decision and Analysis Center processes this information to analyze options and select the most optimal result according to the user's request.
• Management Layer: This class manages the operation of the smart parking system.
Parking Services
4.3.1 Searching Service and Booking Service
1 Searching Service based on Best Suggestion
Users can select a parking slot based on:
• availability of free car parking spaces and a high probability of successfully parking
• optimal solution (calculated based on distance, cost, and total number of free slots)
The parking selection system provides users with multiple options to find the most suitable parking space by considering key factors such as shortest distance, lowest price, and maximum vacancy Users can customize their preferences by adjusting the weights α, β, and γ within the range [0, 1], enabling personalized recommendations based on either a single criterion or a combined approach This flexible approach ensures a comprehensive and user-centric parking assistance experience.
Our parking reservation software for Android smartphones offers users three convenient options: select the nearest parking lot based on distance, choose the most cost-effective parking spot that balances fee and proximity, or customize their search preferences to find the optimal parking solution.
39 optimization of three factors: the parking fee, the optimal distance, and the total number of free car parking spaces
Users must specify their arrival time when reserving a parking space, enabling the system to select the most suitable lot and provide optimal results The reservation is only valid if the user arrives at the designated time; otherwise, it will be canceled If the reservation period expires without the user entering the parking lot, the system sends a notification to prompt a renewal; failure to renew will free up the parking space The software interface visually indicates parking space statuses: red for occupied, green for available, and yellow for reserved As shown in Fig 4.2, the interface displays options for reserving a parking space within the next 30 minutes, facilitating seamless parking management and reservation accuracy.
Figure 4 2 Illustrate the Searching/Booking Service
The tracking service in the diagnostic module allows users to monitor their real-time location on the route to the parking lot and estimates their travel time, providing valuable feedback As shown in Fig 4.3, users can view a map displaying the locations of nearby parking lots and other system users, helping them select parking options with less traffic This feature enhances the likelihood of finding available parking spaces and aids in distributing parking demand across the network, preventing bottlenecks and improving overall traffic flow around parking areas.
Figure 4 3 Illustrate the Tracking service
The guidance service is introduced at the bottom of Fig 4.4
There are three phases of our guidance service:
The outdoor navigation system guides users to their parking spots after booking, enhancing convenience and efficiency Based on Google Maps, it provides detailed directions supplemented with specific information about nearby car parks identified through the user's GPS location This allows users to quickly find suitable parking options and save time, reflecting the effectiveness of our approach, as demonstrated in our previous work in [41].
During the parking entry process, a mini map of external directions is sent to the user's smartphone, displaying their current location and guiding them toward the garage gate for authentication.
The indoor guidance system assists drivers in navigating to their designated parking spaces efficiently Using the integrated mini-map alongside on-site directional instructions, users can easily find their way within the parking facility This seamless navigation enhances the overall parking experience, making it quicker and more convenient for drivers to reach their desired spot.
Figure 4 4 Illustrate the guidance services: (a) Outside guidance, (b) Entering car park guidance, (c) Indoor guidance
4.3.4 Locking Service and Warning Service
A modern RFID-based locking service allows users to securely lock their vehicles in a fixed location while tracking the duration of parking This system records the exact time when the car is first locked, providing accurate monitoring and accountability To exit the parking area, users simply scan their RFID tags to unlock their vehicle, ensuring a seamless and secure process This technology enhances parking management by combining convenience with real-time tracking and access control.
The warning service alerts users through their smartphones when they are approaching an already occupied parking space, helping them find available spots more efficiently It also notifies users if they have exceeded their allocated parking time, preventing parking violations This service is ideal for individuals who need to park within specific time limits, ensuring they adhere to designated parking durations Overall, the system enhances parking management by providing real-time alerts, improving convenience, and promoting responsible parking behavior.
Figure 4 5 Illustrate the Car Locking and Warning Service
Implementation
We developed the software application for our parking system based on the Android platform, ensuring user-friendly interface and seamless connectivity [23-24] The system allows users to log in, search for available parking spaces, and receive recommendations for the most cost-effective options based on waiting time and expenses, optimizing user convenience To assist drivers in locating their reserved parking spots, we offer comprehensive guidance services structured into two phases: outside guidance and indoor guidance, as illustrated in Figures 4.7 and 4.8 This integrated approach enhances parking efficiency and user experience within the system.
For users seeking to reserve a parking space near FCU University, there are two nearby options: the FCU Parking Lot, offering 34 free spaces, and Fuxing Parking Lot, with 24 free spaces Based on our optimized parking selection algorithm detailed in Chapter 3, users are recommended to choose the FCU Parking Lot, as illustrated in Figure 4.6.
Figure 4 6 Procedures reserved parking spot
Table 4.1 provides essential reference information for each user, including reserved parking spaces and current status When the navigation system is enabled, it determines whether the user is inside or outside the parking lot, utilizing different technologies for guidance Inside the carpark, an iBeacon-based system is used for navigation, while GPS is employed for outside guidance The reference table features key parameters such as the NumberRS column, indicating the number of reserved parking spaces, the lock_state column, which shows whether the reserved space is occupied (1) or available (0), and the car_state column, revealing if the vehicle is inside the reserved parking area (1) or outside (0).
Table 4 1 Reference information for each user
NumberRS lock_state car_state
Our GPS navigation system is designed for outdoor carpark navigation, utilizing key functionalities such as transmitting the destination to Google’s API to obtain route data The system receives location points at each corner along the route, which are connected to create an accurate navigation path These location points are continuously updated every few seconds to ensure real-time guidance The detailed implementation interface is illustrated in Fig 4.7, providing a comprehensive overview of the system’s operation and data flow.
Figure 4 7 Implementation of outside car park navigation system
We proposed using iBeacon messages for inside navigation system When the driver goes into the reserved carpark, its GPS signal strength and the number of satellites
The application on the handheld device automatically activates Bluetooth to scan nearby iBeacon messages and accurately determine the driver’s current location using RSSI signal strength This enables the system to provide indoor maps based on the driver’s position and real-time updates of the relative location between the parking spot and the driver, helping them reach their desired parking space efficiently The internal navigation system is illustrated in Fig 4.8.
Figure 4 8 Implementation of inside carpark navigation system
To detect a user’s location, the implementation steps are as follows:
Step 1: An occupied parking slot is assigned
Step 2: Detect all available beacons The beacons will be sorted according to signal strength The strongest beacon signal indicates the user’s current location
Step 3: Each beacon serves as a location point; the system will connect the points to form a navigation road which leads to the destination slot
Step 4: Repeat step 2 and step 3 until user exit the system
To ensure effective beacon interference, the minimum distance between beacons should be at least 7 meters, preventing signal overlap, while maintaining a maximum distance of 40 meters from user devices to ensure reliable connections Proper placement within these ranges optimizes beacon performance and enhances user experience.
Enhance the Performance of Wireless Communication in
Indoor guidance using Ultra-wideband signals
Real-time indoor positioning systems are rapidly developing and are essential for various indoor applications, especially in Indoor Positioning and Asset Systems (IPAS) Traditional GPS systems cannot ensure sufficient accuracy indoors due to signal attenuation and an accuracy of only within 10 meters, making them unsuitable for indoor object localization Therefore, new technologies are needed to improve indoor positioning accuracy This thesis proposes using Ultra-Wideband (UWB) radio signals for high-precision object localization, significantly enhancing indoor guidance services for intelligent parking systems UWB-based positioning methods offer high accuracy and reduce errors caused by timing deviations in both transmission and reception, addressing the limitations of conventional systems.
UWB technology offers a new and more efficient solution for Indoor Positioning and Asset Tracking Systems (IPAS), outperforming traditional methods such as Bluetooth, Cellular, RFID, Image-based technologies, and Dead Reckoning, which are prone to obstacles, multipath interference, and noise Unlike infra-red and ultrasound sensors, UWB does not require line-of-sight and remains unaffected by external disturbances due to its high bandwidth and robust signal modulation Additionally, UWB equipment is cost-effective and consumes less power, making it suitable for large-scale parking lot applications In an IPAS scenario, the system involves installing anchor nodes throughout the parking area and equipping vehicles with UWB tags, enabling precise indoor navigation for drivers after entering the parking lot, helping them locate their reserved spots efficiently.
Figure 5 1 Enhance the accuracy of indoor navigation method for intelligent parking system using UWB signals
UWB technology utilizes various positioning algorithms, including time of flight (TOF) estimation methods such as time of arrival (TOA) and time difference of arrival (TDOA), as well as angle of arrival and received signal strength localization techniques However, these methods often face challenges related to synchronizing time between anchors, which can affect accuracy To address this issue, this thesis proposes employing the two-way ranging (TWR) method for more reliable TOA estimation, enhancing the overall precision of UWB-based localization systems.
The basic Two-Way Ranging (TWR) procedure involves a communication process between the anchor and tag, as illustrated in Fig 5.2 To measure distance, two message exchanges are necessary The tag initiates TWR by sending a Poll message to all anchors in the testing area at time T_SP The anchor records the reception time as T_RP and responds with a Response message at time T_SR, which includes the message ID, T_RP, and T_SR The tag then receives this Response message and records the reception time as T_RR, enabling accurate distance measurement through these timestamps.
The TSP, TRR, TRP, and TSR measurements are used to estimate the Time of Flight (TOF), which in turn allows for calculating the distance to each individual anchor Based on these distance estimates, the current position of the vehicle within the testing room is determined using equation (5.3) This method provides precise localization by integrating sensor data and mathematical modeling.
Figure 5 2 Estimate the TOF between vehicle’s tag and anchor
From this figure, we can estimate the distance and TOF such as: c TOF est distance where c is speed of light ( c 3 10 8 m/s) and
Let S(X,Y) is the current position of the user in the parking lot Figure 5.2 describes the method for determining the user’s coordinate in the parking lot
Figure 5 3 Determine the user’s position in the parking system
As shown in Fig 5.3, we can calculate the current location of the user S(X,Y) using the Pythagorean theorem in triangular
X D Y X S where: D1, D2, D3 and D4 are the relative distance between the vehicle tag and Anchors And, these values are estimated based on two-way ranging TOA as mentioned in above
To evaluate the performance of proposed method, we carried out an efficiency comparison between our test results and the results of related work done by Chang et al
Our proposed indoor positioning method demonstrates high accuracy, with over 53.85% of the measurements having an error within ±10 cm, 86.10% within ±20 cm, and up to 94.85% within ±30 cm, as shown in Fig 5.4 Compared to recent related work, our approach exhibits superior filtering efficiency and positioning precision In contrast, Lin et al.’s method performs poorly due to larger errors exceeding 1 meter, low beacon density, and reduced accuracy. -Boost your article’s SEO and coherence with expert sentence rewriting tailored to indoor positioning insights—see precision in every word!
Positioning accuracy varies depending on the number of beacons used; while Rida et al.’s method, which utilizes three beacons, offers greater precision than single-beacon approaches, it still faces errors up to 1 meter Chang et al.’s method outperforms those of Lin et al and Rida et al by effectively filtering RSSI signal deviations, resulting in higher accuracy However, beacon-based positioning systems have limitations, including a short operational range of 20 to 50 meters and reduced reliability in unstable or industrial signal environments Additionally, beacon messaging frequency depends on the device and can impact performance Consequently, our proposed method consistently delivers superior positioning performance across various environments.
Figure 5 4 Efficiency comparison between proposed method and other related works
Improve the quality of data transmission in IPAS system using a look up table
In an IoT-based intelligent parking system, most sensors are designed for low power consumption and are powered by batteries, ensuring efficient long-term operation These sensors regularly transmit their monitoring data to the central control unit, as illustrated in Figure 5.5 The data transmission occurs within an indoor wireless environment that is prone to high bit error rates, highlighting the importance of reliable communication protocols to maintain system accuracy and performance.
High Bit Error Rate (BER) leads to significant packet loss, necessitating multiple retransmissions that consume substantial energy from network nodes, especially those powered by batteries, thereby reducing system performance and accuracy In parking lot sensor networks, effective communication also occurs between users’ devices and control units to deliver navigation data, emphasizing the need for energy-efficient and accurate data transmission To address these challenges, this thesis proposes employing a Lookup Table Structure for ECC coding selection in wireless communication, aiming to improve accuracy while minimizing energy consumption in IoT devices.
Figure 5 5 Wireless communications in a local parking lot
In wireless communication, packet loss due to errors can be addressed using two primary methods: Automatic Repeat re quest (ARQ) and Forward Error Correction (FEC) ARQ involves adding parity-check bits to each packet before transmission, allowing the receiver to verify packet integrity upon arrival If errors are detected—indicating packet corruption—the system requests a retransmission High bit error rates (BER) in the channel lead to increased packet errors and consequently, more retransmissions, impacting overall communication efficiency.
Low energy efficiency in data transmission occurs when more energy is required to successfully transmit corrupted packets Forward Error Correction (FEC) enhances data reliability by encoding and decoding data packets using various channel coding techniques Among these, Reed-Solomon codes, represented as RS(n,k), are the most widely used method to detect and correct errors during transmission, improving overall communication robustness.
This thesis explores the implementation of advanced coding techniques utilizing two lookup table structures for efficient data transmission in IoT devices within IPAS systems The proposed approach involves deploying two lookup tables at the sender node, enhancing data accuracy and transmission reliability These innovative coding methods aim to optimize IoT communication performance by leveraging structured data mappings, as outlined in our previous study.
The first table presents the optimal Reed-Solomon (RS) code values tailored for various communication distances An optimal RS code is defined as one that minimizes recovery overhead while maximizing energy efficiency during data transmission Crucially, there is a specific optimal code for each distance, with shorter code lengths being preferable for shorter ranges and longer codes suited for extended distances This table guides the initial transmission, ensuring the most efficient and reliable data transfer based on the distance.
The second table provides the optimal RS(n, k) code values tailored to various BER levels of the real communication channel Since channel conditions fluctuate—such as increased BER when a mobile node moves indoors—a static code selection becomes inaccurate To address this, the second table enables dynamic adjustment by offering optimal RS codes based on current channel performance, ensuring reliable communication despite changing conditions.
To select the optimal Reed-Solomon (RS) code, we base our decision on the feedback message received after the initial transmission This feedback provides valuable insights for choosing the most effective code for subsequent transmissions Our algorithm performs multiple transmissions, utilizing the feedback results to dynamically adapt and optimize the RS code, thereby improving overall data reliability and transmission efficiency.
• At the first transmission: Based on the distance that has been pre- calculated, the sender node will select an optimal value of RS(n, k) code in the distance table
After the initial transmission, we assess the channel conditions, including the Bit Error Rate (BER), and select an optimal RS (n, k) code for subsequent transmissions based on BER data, ensuring improved communication reliability.
• At the third to n-th transmissions: Based on the result of the previous transmission, we will increase or decrease the length of the redundant bits in the
RS code to achieve better performance (lower overhead and better energy efficiency) The selection is based on second table
The detailed structure of these table was introduced in [41], which is showed a high energy efficiency and significant reduction of packet error rate
Figure 5 6 Apply the Lookup table structure to minimize the overhead of data transmission in IPAS system
This study simulates the selection of RS codes in the proposed algorithm by varying the transmission distance between the control unit and the vehicle to identify the optimal RS code length for each location Using eight bits per symbol encoding with k = 233 bytes, a total of 16 RS codes are evaluated For each fixed transmission distance, the RS code with the best performance—termed the optimal code—is selected Since the channel's bit error rate (BER) fluctuates, selecting RS codes solely based on distance is inadequate To adapt to real-time channel conditions, a second table is constructed to store optimal RS code lengths corresponding to different BER levels, enhancing the robustness of the coding strategy.
Our proposed algorithm selects the optimal RS code for subsequent transmissions by analyzing the actual channel bit error rate received through ACK messages This approach allows us to determine the RS code that minimizes recovery overhead, ensuring efficient data transmission As shown in Figure 5.7 (a), we compare the recovery overhead of various codes across transmission distances of 10, 20, 30, 40, and 50 units, highlighting the importance of adaptive code selection for improving communication efficiency.
60, 70, and 80 m Figure 5.7 (b) shows the comparison of the recovery overhead of all the codes when the bit error rate changes
Figure 5 7 Comparison of Recovery Overhead of different RS codes: (a)
Communication distance changes; (b) BER changes (with k = 223 bytes)
To evaluate our proposed algorithm's performance, we compared it with methods from references [57] and [58], focusing on average recovery overhead when the number of transmissions (t) is 6 Our Adaptive Lookup Table (ALT) method consistently achieves the lowest recovery overhead across various scenarios, with particularly significant reductions at large transmission distances (> 20 m) The algorithm in [58] outperforms [57] by acquiring channel characteristics such as BER and RSSI after the first transmission, enabling it to select more efficient coding and reduce transmission overhead In contrast, the algorithm in [57] performs the worst, as it requires multiple transmissions to determine the optimal Reed-Solomon code length, resulting in slower response to changing channel conditions and higher overhead.
56 of transmissions required to find the optimal RS code is longer than that required by the other algorithms
Figure 5 8 Comparison of average recovery overhead of proposed Adaptive Lookup-Table method, AMFEC [57], and Ghaida [58] with number of data transmissions t = 6
An adaptive solution for data transmission between local IoT network and server
For intelligent parking system that used model architecture as shown in Fig 5.9, we propose a new method to reduce the end-to-end delay of real-time applications when the number of sensor nodes increase The use of MQTT protocol in sensor networks (MQTT-SN) plays an important role in IoT applications MQTT is a lightweight protocol for applications in which the IoT nodes have a limited energy supply But, the implementation of this protocol faces many challenges in terms of the reliability of real- time application systems Some previous studies in [45, 64] have explored and analyzed the parameters affecting an end-to-end system using the MQTT protocol They consider parameters such as the probability of and the delay in content delivery within a
57 healthcare, image streaming systems using IoT technology They also propose a mathematical model for estimating the end-to-end service delay and the probability of content delivery However, these approaches have disadvantages of a large end-to-end delay or congestion when the number of network nodes increase
Figure 5 9 A new architecture for real-time data transmission in IPAS system using
MQTT protocol Figure 5.9 describes the architecture of the proposed system We designed a parking system that can monitor parking information using sensors (or cameras) [42, 44], this information includes: occupied status of each parking space in car park, live image of each parking space (if using camera nodes), and information of surrounding environment of each parking space Our system help transmits data of car parks in real-time with less end-to-end delay when the number of sensor nodes increase
The system comprises local MQTT client nodes installed in a car park, an MQTT broker (gateway), and a cloud server to facilitate seamless data transmission Each local MQTT client connects to sensors or cameras positioned at each parking space—six cameras were used in our experiments—capturing environmental data These MQTT clients establish Wi-Fi connections to the MQTT broker and subscribe to specific topic IDs, such as parking system/car park-ip/parking space-ip, where each parking space and MQTT client have unique identifiers The MQTT clients continuously send data to their respective subscribed topics, enabling real-time monitoring and management of parking spaces through an efficient IoT-based solution.
The MQTT broker functions as a gateway in this system, receiving real-time data, including images and data speed information, from local nodes and forwarding it to a cloud server A user's smartphone with an internet connection, such as 4G, is required to access the real-time data The gateway dynamically adjusts the upload speeds of connected nodes by analyzing their data rates and comparing them to predefined thresholds If the total data speed exceeds the threshold, the gateway sends a 'publish' message to subscribed topics across all local nodes, prompting immediate data rate adjustments for optimal system performance.
End to end delay estimation
An important parameter to evaluate the performance of proposed IPAS system is end-to-end delay The equation for computing the end-to-end delay can be calculated as:
In MQTT communication, the round-trip TCP delay is determined by D CONNECT − TCP, D SUB − TCP, and D PUB − TCP, which represent different stages of the protocol The network connections involved include N-G (sensor node to gateway via Wi-Fi), G-S (gateway to server via 4G), and S-U (server to user via 4G) When data is transmitted using the HTTP protocol, the total round-trip TCP delay is represented by D REQ S − U − RES, illustrating the time taken for requests and responses between the server and user These latency measurements are essential for optimizing IoT and web communication performance.
Figure 5.10 demonstrates how queue length influences the performance of the proposed system, highlighting the differences between a standard network and the adaptive network The comparison shows that the adaptive network effectively manages varying numbers of sensor nodes, especially as the queue length increases from 30 to higher values, resulting in improved performance and reduced delays This analysis underscores the importance of implementing adaptive strategies to optimize network efficiency under different traffic conditions.
60 Figure 5.10 (a) shows a comparison of the value of the end-to-end delay of the normal network and the proposed network when the number of sensor nodes is increased from one to six We can see that when the number of sensor nodes is small, the end-to-end delay values for the proposed network and the normal networks are approximately the
Increasing the number of nodes in the network results in higher traffic and more messages exchanged, but the proposed IPAS network significantly reduces end-to-end delay Analysis shown in Figure 5.10(b) indicates that the proposed IPAS network maintains a consistently lower packet drop rate compared to the standard IPAS network, ensuring improved reliability and performance.
Figure 5 10 Comparison between the adaptive network and normal network for changes in queue length and number of sensor nodes: (a) end-to-end delay; (b) packet drop rate