An efficient node selection algorithm in the context of IoT-based vehicular ad hoc network for emergency service

1 Introduction

Vehicular ad hoc networks (VANETs) are an essential technology in the deployment of the intelligent transportation system (ITS), which aims to improve road safety and introduce environmentally friendly driving. VANETs can establish ad hoc communications with their peers or with infrastructure networks by vehicles with onboard wireless communication facilities [1]. Recent breakthroughs in sensor and network technology have made the pervasive enhancement of the ITS possible. ITSs are management systems that improve transportation efficiency by integrating information, computer technologies, sensors, and communication [2,3]. The ITS uses roadside units (RSUs), with vehicles acting as mobile sensors, such as smartphones or in-vehicle technologies. This system monitors the surrounding environment and traffic conditions [4,5]. Research on VANETs is an effective way to assist the ITS in the rapprochement of VANETs and intelligent terminal technology. VANET nodes are vehicles equipped with processing units, embedded sensors, and wireless interfaces to support infrastructure communication [6,7]. These communications can improve accident prevention applications and safety on roads, as well as reduce congestion [8]. Additionally, it utilizes available road capacity to propose less congested shortest routes for drivers, which directly translates to reduced commute times for travelers [9,10,11]. Typically, the vehicle traffic routing system determines a path between two points by considering various criteria. Moreover, different adaptive traffic light systems, functioning through wireless communication between vehicles and stationary RSUs, have been implemented and investigated to observe and control traffic congestion situations at conjunctions [12]. It must be noted that vehicle traffic routing systems constitute the primary focus and concern of this research, which is to find the optimal route for the ambulance to reach the patient and provide appropriate aid.

To enhance VANETs’ performance and connectivity, mitigate traffic issues, and decrease road accidents, the Internet of Things (IoT) is employed to address various challenges in the urban traffic environment [13]. Furthermore, it provides network access and travel schedules to drivers, passengers, and employees of the traffic organization division. The VANET based on the IoT establishes a connection between diverse road networks and wireless technologies [14,15]. This advancement in ITS is used to improve the intelligent healthcare system, protect people from traffic accidents, and build a traffic monitoring system [16,17,18,19]. This work utilizes this technology to develop a revolutionary node rank index (NR-index) technique for choosing the suitable ambulance with the optimal route to reach the patient.

One of the most significant issues facing emergency medical services (EMS) is the absence of multi-stakeholder coordination [20]. Particularly in the process of routing ambulances, unresponsive traffic operators need to coordinate effectively in selecting the appropriate ambulance vehicle when employing the greedy-based selection algorithm. However, the existing node selection algorithms endeavor to solve the limitations of classical greedy-based selection by involving different metrics in their decisions. They have no reflection for road configuration, traffic density, vehicle speed, or traffic awareness metrics [21,22,23].

Thus, the current research contributes toward alleviating the previously mentioned concerns by proposing a real-time NR-index algorithm, which selects the best node (ambulance) to supply medical services to emergency cases by involving traffic awareness metrics like congestion level and distance in the node selection decisions. Basically, the proposed algorithm finds the best route to the destination while avoiding traffic congestion. A common system was designed to supply medical services for emergency cases and to help ambulance vehicles avoid overcrowded zones within an ITS. First, the proposed system uses public transportation systems and VANETs to achieve effective real-time communication between the emergency center, RSUs, hospitals, and vehicles (e.g., ambulances). Next, a real-time NR-index algorithm was proposed to find the best ambulance to provide emergency medical assistance and plan routes to it while reducing the traveling cost through the ability of vehicles to avoid congested areas. The significant contributions of this work are summarized as follows:

• This article focuses on the ambulance routing problem in VANETs by considering multiple factors, such as distance and traffic conditions.

• Introduce weights for different factors to compute the route from patients to hospitals with the shortest and least congested paths.

• The real-time NR-index algorithm is developed for ambulance vehicle routing based on traffic conditions to reduce transport time and journey cost.

The structure of the article is organized as follows: The related works on routing algorithms, particularly from the perspective of traffic management strategies based on priorities, are presented in the next section. Following that, we introduce the IoT emergency services system based on the proposed NR-index algorithm. We will then give the simulation experiments and performance evaluation of our proposed work. Finally, conclusions and future work will follow in the last section.

2 Related works

The objective of the ambulance routing problem is to locate the most efficient route for both picking up and dropping off casualties within a given network [24,25,26,27]. The different models formulate both the actual road network and the traffic conditions [28]. Jotshi et al. [29] suggested an approach to finding the shortest path while taking hospital availability, distance, clustering criteria, and patient priorities into account. The proposed approach used data fusion to estimate the number of casualties, damaged infrastructure, and road conditions in a post-disaster environment. Moreover, the introduction of network partitioning in this approach aims to reduce computational complexity.

Zeng et al. [30] focused on investigating the ambulance routing problem involving multi-stakeholder cooperation. Traffic operators focus on offering traffic-controlling approaches to pre-clear roads and rank the inbound ambulance. Meanwhile, hospitals carry out distant screenings to make sure they have the requisite expertise to treat arriving patient(s), should they be assigned to handle the emergency. A mixed-integer linear programming model was planned to reduce expenses for ambulance allocation, traveling, and delay penalties. The decisions of EMS on the vocations of hospitals, as well as ambulance types and fleet size, can be reinforced by employing the proposed technique. The particular solution of the utilized optimization model takes only a few seconds to complete, which facilitates stakeholder coordination within a short response interval. Accordingly, the forwarded model and feasible rescue strategies can capably handle varying emergencies. The model can likewise provide dispatch options for the heterogeneous ambulance fleet and match patients to the most suitable vehicle to prevent the wastage of limited resources. Although this model can solve daily ambulance routing issues, it creates a negative impact on general traffic status.

Qing et al. [31] utilized a well-known greedy routing technique (Dijkstra algorithm) that can efficiently determine the shortest route between a reference node and the further nodes in the chart. This algorithm exploits a greedy tool to discover the straight path from the source node to the endpoint. In each step, the lightweight route is chosen. Dijkstra’s search technique can thus be seen as an escalating ring around the source node that keeps going in every direction to reach the target. However, the improved Dijkstra algorithm proposed in this study tries to find and evaluate all possible shortest paths, but it can only store data for previously visited nodes.

Greedy perimeter coordinator routing (GPCR) is another greedy technique that routes packets between street intersections using the greedy forwarding strategy and delegates routing decision authority to the nodes at the intersections [32]. Nevertheless, the main drawback associated with GPCR is that it is not a traffic awareness technique, and it does not consider link connectivity when determining the optimal path. An enhanced version of the greedy traffic-aware routing protocol (GyTAR) was introduced by Jerbi et al. [33]. It utilizes geographic intersection data to identify reliable and efficient routes in urban environments. The GyTAR approach holds the nodes (vehicles) information in the routing table and utilizes the traffic density as an evaluation factor, which is very bandwidth-intensive.

SmithaShekar et al. [34] created an effective navigation system for ambulances founded on VANET. By using historical data and real-time traffic announcement improvements, the system aims to prevent unexpected congestion while tackling the problem of finding the shortest route to the destination. The proposed dynamic routing system uses global positioning system (GPS) data and traffic situations in a timely manner. This system, which combines a road transport system and a metro rail network, is designed to guide ambulances in real-time situations and dynamically adjusts traffic lights. Similarly, Djahel et al. [35] have presented a comparable adaptive system to improve emergency vehicle traffic management. This framework suggests changing driving policies, modifying driver behavior, and implementing suitable security measures. Nonetheless, further assessment is required for the validation and performance evaluation of the proposed framework.

Sundar et al. [36] developed an intelligent traffic management system to allow ambulances to pass and control traffic congestion. Radiofrequency identification tags are attached to vehicles to determine the number of vehicles on specific roads, diagnose stolen vehicles, and send alerts to the control room. In addition, the system prioritizes ambulances by cooperating with traffic controllers using ZigBee modules. To implement this system, an intelligent road infrastructure would have to be distributed across multiple roads, and this is the main restriction.

In a previous study [37], integrated cellular technology is suggested to be used in the transportation industry as an accident management system. Components in ambulances, RSUs, and servers can all able to connect with each other. Additionally, to optimize the increasing spatial utilization of road networks and lower vehicle operating costs, an optimal route-planning algorithm (ORPA) is proposed for this system. It is also possible to use the suggested method to give ambulances quick routes through congested areas. All cars, even ambulances, have to have remote connection capability and a route indicator installed. The assessment information showed that the ORPA operated effectively in terms of average speed and travel duration. Because of the unpredictable nature of traffic, the proposed method works best for expected patterns only.

3 Proposed IOT emergency services system

This section investigates the relationship between traffic conditions and expected times for emergency response as a methodology of the developed emergency services system. This system is an IoT emergency services system based on the developed NR-index algorithm to provide medical services to emergencies by finding the best ambulance. The developed algorithm presents a lightweight route from the emergency response center to the patient and then to the hospital. This algorithm avoids traffic congestion by using distance and congestion level factors to discover the best travel route to the destination. The proposed IoT emergency services system includes emergency centers, medical IoT devices, patients, ambulance vehicles, and RSUs. Finally, each system component is part of the network in this work.

In the proposed IoT system, each road within the network coverage is divided into segments. The RSUs are fixed at the beginning and end of each segment. An identification number (ID) is assigned for each RSU, road segment, and vehicle. The developed NR-index algorithm will draw a speed roadmap by utilizing the received information regarding the average speed of the vehicles in road segments from the distributed RSUs. The NR-index algorithm has connections to all the entity units and manages the speed map module using the average speed of vehicles. Moreover, it is also responsible for computing the lightweight route from the closest ambulance vehicles toward the patient. It utilized the speed roadmap and the distance to choose the appropriate ambulance to handle the emergency case by sending the ambulance (ID) and the lightweight route information to the emergency response center. The ambulance then uses this information with the assistance of the GPS to identify the journey route. The proposed system consists of two primary phases: the emergency alert detection phase and the emergency response phase. The details of both phases are as follows:

3.2 Emergency medical response phase

Once the emergency center receives the alert message, the candidate ambulance vehicles within the range of the patients are determined. Then, the roadmap, which was created using information gathered from the distributed RSUs, is utilized to find the low-cost journey. Depending on the distance and congestion level weights, the proposed NR-index algorithm computes the NR index for each node (ambulance) and chooses the most suitable ambulance vehicle to respond to the emergency case through a low-cost journey. For this purpose, this article suggests two parameters for the trade-off between congestion and distance factors. The first parameter is the distance between the patient and the surrounding ambulance vehicles through the distance formula, which is given by

(1) Dist ( A , P ) = ∑ e i = 1 ( S _ dis ) 2 + ( ℓ _ int ( i ) 2 − ℓ ′ _ int ( i + 1 ) 2 ) ( S _ dis ) 2 ,

where (S_dis) is the shortest distance between the patient and the candidate ambulance vehicle, as shown in the red dotted line in Figure 1. ℓ is the Euclidean distance between the ambulance and the next eth intersection toward the patient. In contrast, ℓ ′ is the distance that separates the successive intersections to the patient.

Figure 1

Conceptual architecture scenario.

After identifying the candidate ambulances based on the distance parameter, the NR-index algorithm calculates the road segment average speed to draw the road map speed, which can be calculated as

(2) S avg = ∑ m = 1 k ∑ n = 1 l R s   V n N ,

where R _s is the road speed limit and V _n is the speed of the nth vehicle in a specific road segment mth. Meanwhile, N denotes the vehicle number in the road segment. Based on the road part average speed ( S avg ), the proposed algorithm can determine the congestion level and avoid the path with high congestion traffic in the transportation network. The proposed IoT emergency services system is represented as a workflow diagram, as shown in Figure 2, and the components are defined as activities or services that work on a cloud infrastructure.

Figure 2

Workflow of the proposed IoT emergency services system.

Equation (3) develops the NR-index calculation to examine the values of the weighting factors integrated into the NR-index function. It incorporates the value of Dist and S _avg of the evaluated path based on the weights of β ₁ and β ₂

(3) Node rank index = β 1 ⋅ Dist + β 2 ⋅ S avg .

The weighting factors are mainly employed to specify each component’s percentage in estimating the NR index. The total summation of the contributions ratios that β ₁ and β ₂ and must equal 100% (i.e., β ₁ + β ₂ = 1) [38]. Generally, there are no optimum ratios to be apportioned to β ₁ and β ₂, which contribute to the finest performance in various situations. It is significant to indicate that the delicate balance between Dist and S _avg is essential to the overall effectiveness of the suggested NR-index algorithm. Assigning equivalent weight for both factors results in a more balanced effect of Dist and S _avg in different vehicular density environments. Moreover, giving equal importance to both congestion level and shortest path factors can contribute to optimizing node selection choices.

Therefore, assuming the values β ₁ = 0.5 and β ₂ = 0.5 indicates that each of Dist and S _avg resulted by 50% in the calculation of the NR index. Algorithm 1 illustrates the whole working principle and logical sequence of the proposed IoT emergency services system.

Algorithm 1: Proposed IoT Emergency Services System

Step 1: Initialization: Attach biosensors to patients’ body to collect biometric data

Step 2: Continuous monitoring of patient’s vital signs or physiological functions

Step 3: If the body biometric measurements exceed the average rate, then

Step 3.1: A detector sensor prepares an alert message containing (clinical value. patient location).

Step 3.2: Send an alert message to the emergency center.

Step 4: End if

Step 5: Else go to step 2

Step 6: The emergency center finds the set of ambulance vehicles (A) within the range of the patient’s location

Step 7: For each a ∈ A

Step 7.1: The proposed NR-index algorithm computes the node rank index for each a ∈   A;

Step 7.2: The NR-index algorithm maintains the priority table by inserting (ambulance (ID), node rank index);

Step 7.3: Sort the priority table according to node rank index (lowest value first);

Step 7.4: NR-index algorithm forwarding the alert message to the candidate ambulance;

Step 8: End For;

Step 9: The ambulance vehicle uses the information contained in the alert message to provide emergency medical care for patients in a timely manner.

Step 10: End.

4 Simulation experiments and performance evaluation

The proposed IoT emergency services system was simulated and evaluated using MATLAB/Simulink as the simulation environment. The general Manhattan grid model [39] has been utilized with Google Maps to conduct a case study. A number of urban network scenarios were generated to validate the impact of performance parameters on path selection metrics. To enable realistic traffic scenarios, the simulation utilized the mobility models with the ability to create the most common information regarding the vehicle’s acceleration, road direction, traffic lights, and edges. The simulation’s parameters are shown in Table 1.

In this section, the performance of the proposed NR-index algorithm is assessed and evaluated against the well-known Dijkstra algorithm according to the average travel time (ATT), average travel speed (ATS), and normalized routing load (NRL).

Two scenarios are applied to assess the performance of the proposed NR-index algorithm in selecting the best ambulance vehicle to respond to emergency cases over the low-cost journey in urban environments. In the first scenario, vehicle density is a crucial point in simulation settings, and it reasonably affects the precision of the received results. The second scenario was applied to assess the proposed algorithm’s behavior as compared with the other algorithm when a road accident occurs. To accomplish the second scenario, 300 vehicles were simulated, and then the number of accident occurrences from 1 to 5 was randomly placed in the simulation area.

4.1 ATT

Figures 3 and 4 illustrate the results for the proposed NR-index algorithm and Dijkstra mechanism in terms of ATT, which is the duration spent by a particular vehicle to reach its destinations from the start point. The outcome was investigated in relation to the two specified scenarios, namely, different vehicle densities and accident conditions. According to the results under different vehicle densities (Figure 3), the ATT of a vehicle is directly connected with vehicle density; that is, the ATT increases as the number of vehicles rises. However, this rise is too intense when depending on the Dijkstra mechanism. This is because the Dijkstra mechanism guides the vehicles via the direct path only without awareness of other factors such as vehicle congestion and speed. By comparison, the NR-index algorithm significantly improved the mean of travel time as compared with the Dijkstra mechanism. The ATT of the NR-index and Dijkstra algorithm is almost the same at low compound densities (100–200). At the same time, the ATT value increases at higher densities. The NR-index algorithm has the finest influences in various densities, and it reduced ATT by up to 17% compared with Dijkstra. This enhancement in ATT is due to the NR-index algorithm’s ability to avoid congestion rather than encounter it, which is not considered in the Dijkstra algorithm.

Figure 3

ATT for varying numbers of vehicles.

Figure 4

ATT for varying numbers of accidents.

In Figure 4, the routes are directed over the shortest and straight path by the Dijkstra algorithm. However, after the congestion occurred on the path due to an accident, Dijkstra ignored this matter and continued to route the ambulance vehicles through the shortest path without assuming the accident’s stringency, duration, and condition. To relieve this weakness and to assume that accidents occur implicitly, the NR-index algorithm guided the vehicles via the possible least congested and shortest paths by considering the distance and vehicle speed factors. In Dijkstra, this did not occur due to the need for greater awareness of the dynamic evolutions in the vehicular domains. Therefore, the travel time began to rise snappily for the vehicles routed through the Dijkstra systems.

4.2 ATS

The ATS of the Dijkstra and the NR-index algorithm at different vehicle densities is clarified in Figure 5. The same outcomes were obtained for all algorithms when there were low vehicle densities (100–200 vehicles) because there was less congestion, and the ambulance vehicle was routed via the same path, which is the shortest path. The proposed NR-index algorithm received the most acceptable ATS with different vehicle densities by choosing the lightweight path to emergency cases with less congestion level. However, as the density of vehicles increased, the ATS was reduced for both compared algorithms. The Dijkstra algorithm rendered the most inadequate ATS, which ranged from 21 to 57 km/h. Dijkstra performed the worst when it came to the accident scenario (Figure 6) in terms of ATS because they did not account for the congestion level and routed the ambulance through the shortest path, which was congested as a result of the accident. The ATS of travel was decreased for both approaches after the traffic accidents occurred. However, this decrease is too steep for the Dijkstra algorithm versus our proposed NR-index algorithm. The NR-index algorithm has the best performance because it considers various metrics such as speed and path length simultaneously to choose the appropriate ambulance and then routes it via the lowest-cost path.

Figure 5

ATS for varying numbers of vehicles.

Figure 6

ATS for varying numbers of accidents.

4.3 NRL

This measure reflects the number of congested roadways that the ambulance vehicle might encounter during its journey. In our simulation, a road is considered crowded when vehicles’ speeds are less than 20 km/h during the routing process. Since avoiding traffic jams and finding a lightweight route are essential issues, these parameters are taken into consideration and evaluated to find the best solution. As the outcome, the effectiveness of the Dijkstra mechanism and the suggested NR-index algorithm is evaluated in connection with the number of overloaded roads during an incremental rise in vehicle density (Figure 7). For both approaches, the rate of congested roads rises as the vehicle’s number increases. In the present evaluation, the suggested NR-index performs better than the Dijkstra mechanism and identifies the light route for ambulance vehicles at different vehicle densities. The main reason for the Dijkstra mechanism’s poor performance is that it routes the ambulance down the shortest route without considering traffic conditions.

Figure 7

NRL for varying numbers of vehicles.

Figure 8 shows the numbers of overcrowded roads under accident conditions for both the proposed NR-index algorithm and the Dijkstra mechanism. The number of accidents on roads is consistent across all approaches, as can be observed. However, both methods show an ascending tendency during the simulation period after the accidents occur. Nevertheless, the increase is not substantial in the context of the proposed NR-index algorithm due to its unique routing mechanism, which includes the journey cost from the initial phases of the routing process. In contrast, due to the Dijkstra mechanism’s insistence on directing vehicles along the shortest path without considering congestion level, it performs the poorest in accident scenarios.

Figure 8

NRL for varying numbers of accidents.

5 Conclusions

The dramatic increase in vehicle population rather than traffic calming measures is the most likely reason for the unprecedented challenges that authorities encounter in managing the traffic of ambulances to provide medical services for emergency cases. The time of EMS’ responses is crucial for numerous health outcomes, such as hospitalization, rehabilitation, and survival following a stroke or heart attack. In order to provide effective real-time communication between all entities of the emergency services system, the proposal built an emergency service system that makes use of the networks for IoT healthcare devices and VANETs. The developed emergency service system based on the NR-index algorithm provides a solution to the problems of emergency medical response times by utilizing traffic status while selecting an ambulance vehicle with the path that has the lowest journey cost from the initial stages of the routing process. The NR-index algorithm reduced delays that occur in providing EMS to critical cases due to congestion in the transportation network. Finally, further investigation into the relationship between traffic and emergency response times is needed to develop an enhanced version of the proposed system that will consider different types of priorities in the guidance procedures for various types of emergency vehicles like ambulances, fire trucks, and police vehicles.