https://orcid.org/0009-0002-2578-4377
Figures
Abstract
Emergency Vehicles (EVs) are of considerable significance in saving human lives and property damages. To promote the efficiency of emergency operation, signal preemption control could give priority to EVs heading toward the incident location. On the other hand, providing dynamic and precise route planning for EVs plays an important role in emergency rescue since traffic changes constantly. Furthermore, connected vehicle (CV) technology that incorporates advanced wireless communication technologies, offers a huge potential to promote the efficiency of EVs and maintain smooth traffic flow via collaborative optimization of routes and signals. This study presents a bi-level dynamic emergency route planning system considering signal preemption control, which builds on traffic flow combined with hierarchical bi-layer model predictive control (MPC), for more than one EV under partial CV environment. In this approach, the mobility of EVs is prioritized before decreasing the impact of EVs operation on normal traffic. In the upper layer, a road-level emergency route would be dynamically planned and updated after each time horizon, according to the network-wide traffic flow estimation under diverse CV market penetration ratios through loop detectors and Cellular-Vehicle-to-Everything (C-V2X) communication. In the lower layer, a lane-level emergency route that combined with signal preemption control would be planned to ensure the efficiency of EVs and reduce the adverse impact of signal preemption on normal traffic. In the end, a microscopic simulation environment for a real traffic network is carried out to test the effectiveness of the proposed system. The simulation results indicate that the proposed system provides reliable and practical emergency route planning and signal control services for EVs under different traffic flow conditions.
Citation: Dai Y, Hu L, Zhou S, Liu Y, Yang A (2025) A bi-level dynamic emergency route planning system considering signal preemption control using CV technology. PLoS One 20(5): e0323209. https://doi.org/10.1371/journal.pone.0323209
Editor: Zhihong (Arry) Yao, Southwest Jiaotong University, CHINA
Received: January 19, 2025; Accepted: April 3, 2025; Published: May 19, 2025
Copyright: © 2025 Dai et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Once an unexpected event has occurred, an evacuation or a rescue must be carried out to provide assistance to individuals in hazardous situations, and emergency vehicles (EVs), such as fire trucks and ambulances, play an essential role in this process. Further, dynamic emergency route planning and emergency signal preemption control are two keys reducing casualties and property losses as well as even alleviating road network congestion caused by emergency rescue, which are needed to coordinate work [1–5].
Reasonable arrangement of EV route can shorten the travel time and avoid congestion, thereby the accident loss can be reduced. Regarding the minimum route distance as the target, the traditional algorithms are mainly based on historical data but this might not be always the best option for EVs in practice, since real-world traffic is constantly changing [6,7]. Among the existing studies [8–12], emergency route planning in a road network that considers the dynamic time-varying characteristics of traffic flow has been investigated. They can typically be classified into road-level route planning and lane-level route planning. Compared to road-level planning, lane-level route planning can provide more precise details that are essential for vehicles’ high precision localization and path guidance during driving [13,14]. Furthermore, many studies suggest that lane-level route planning is more efficient and applicable for vehicles possessing communication technology [7, 15–17].
In practice, in spite of dynamic route planning, EVs’ efficiency still might be disturbed by other formal vehicles (i.e., EVs move slowly since the speed of the preceding vehicle is low) and EVs even queue if the subjective signal is red. Hence, in order to guarantee EVs maintain their efficiency, some scholars proposed signal preemption control strategies based on given routes [3,13,18]. The conventional approach favored offering green waves for EVs to arrive at the destination as quickly as possible [19,20]. The wireless sensor network or radio frequency identification detects the emergency vehicles on the roadside intersection, and the traffic lights individually tune its preemption phase. In such kinds of preemption models, the traffic lights adjust their phase after the arrival of emergency vehicles. Nevertheless, emergency signal preemption control usually inevitably interrupts the normal traffic signal timing operations which could even bring superabundant delay to normal traffic [21]. For instance, an existing study [22] shows the advantages of the EVs preemption model at six locations of New York, whereas it demonstrates the disturbances at coordinated signalized intersections caused owing to preemption. Hence, minimizing the travel time without disturbing the normal road traffic is also a major purpose for EVs. Moreover, most of the signal preemption strategies for EVs are activated based on the routes that have been planned [23–26], instead of coordinated working with dynamic route planning for EVs. More specifically, most of the emergency vehicle preemption models are designed for a single emergency vehicle scenario. However, more than one emergency vehicle may present in the same lane, which is a major issue in real-time scenarios.
Dynamic route planning and coordinated signal control can better ensure the efficiency of EVs and maintain smooth traffic flow in the road network. In light of this, our study explores the potential of introducing connected vehicle (CV) technology to deal with this problem. CV technology, which contains advanced wireless communication technologies like Cellular-Vehicle-to-Everything (C-V2X), is valuable and has been gaining attention around the world [27–28]. It has been predicted that the worldwide penetration of CV in new vehicles will increase from 10% in 2018 to 70% in 2027 [29]. Under a CV environment, data is collected via sensors installed on CVs and vehicle state information (i.e., vehicle velocity, position, etc.) are guaranteed to transmit among CVs. Moreover, communication devices equipped on CVs can obtain velocity trajectories suggested by central cloud coordinator, meanwhile, the central cloud coordinator can issue instructions through C-V2X to control signals [30]. Accordingly, CV technology can provide an opportunity to assign space-time priority to EVs via dynamic route planning and signal control [31–33].
To summarize, seldom researchers conduct signal preemption control for EVs combined with dynamic emergency route planning and concerning negative impacts on normal traffic, especially under diverse CV market penetration ratios. Moreover, it is crucial to develop such a system that combines vehicles and infrastructures to improve the efficiency of emergency rescue and reduce the impacts on road networks under partial CV environment. Therefore, the objective of this study is to design a system for EVs that:
- Providing dynamic road-level route planning and lane-level route planning for EV to enhance its efficiency.
- Control signals to provide green wave for EV and reduce the influence of EV operation on normal traffic.
- Taking advantage of CV technology under diverse CV market penetration ratios.
The remainder of the paper is organized as follows: Section II ‘an integrated system of signal preemption control lane-level dynamic emergency route planning’ provides a description of the proposed system; Section III ‘Route planning algorithms and traffic signal preemption control scheme’ presents road-level route planning and lane-level route planning algorithms as well as signal preemption control scheme; Section IV ‘A study case to test the numerical model on a real network’ assesses the proposed system by applying it to a real network; Section V ‘Discussion’ presents this study’s limitations; Section VI ‘Conclusion’ concludes this paper and briefly discusses future research directions.
An integrated system of signal preemption control and lane-level dynamic emergency route planning
In this section, the framework of the proposed integrated system is presented in Fig 1. The system is designed to operate in heterogeneous environments, leveraging existing infrastructure (e.g., loop detectors and mobile signaling data) and optionally utilizing CV technology to enhance real-time coordination between vehicles and traffic lights. Additionally, this structure is based on the assumptions that communication is reliable and there exist no communication issues such as delay and data packet loss.
The system is activated when an EV platoon makes a request. To begin with, an initial route would be generated and we refer to A* algorithm in this step. Additionally, data (i.e., numbers of CVs in the road network, etc.) is collected by C-V2X, which is leveraged to estimate traffic flow of road and lane. Then, estimation results are sent to do dynamic route planning. Due to the complexity of dynamic route planning integrating with signal preemption control, model predictive control (MPC) is introduced to assist in route planning.
As shown in Fig 2, MPC that is an iterative, finite-horizon optimization, is appropriate for dynamic tasks. Specifically, at time t, input state variables are sampled from the past information, and then the expectation of an objective function is minimized in a certain horizon, , to predict an optimal state variables sequence over the horizon. Notably, only the first predicted output state variables are implemented at time t. Then the optimization is sampled again and prediction is repeated with a shifted prediction horizon.
In this study, we design a hierarchical bi-layer MPC for dynamic route planning. The upper-layer MPC undertakes road-level dynamic route planning and provides the lower-layer MPC with selected route segments (i.e., roads and lanes). After that, the lower-lever MPC undertakes road-level dynamic route planning and offers selected lanes to traffic lights, meanwhile, traffic lights provide their signal preemption control schemes to the lower-lever MPC for its iterative prediction. In the end, the selected route that includes selected road and selected lane as well as predicted optimal velocity at time t are supplied for the EVs as the suggested route and velocity. The details of dynamic route planning and signal preemption control are elaborated in the next section.
Route planning algorithms and signal preemption control
This section presents the process of road-level route planning and lane-level route planning as well as signal preemption control in detail. Table 1 lists the indices and parameters utilized hereafter.
Traffic flow estimation
Traffic flow estimation is vital for route planning since it can reflect whether the road capability satisfies the traffic requirement [34]. There are three important traffic parameters: speed, density, and flow. The speed and density can indicate the service quality of the road while the flow stands for how many vehicles are passing the road within a duration. The Fundamental Diagram is a macroscopic traffic model that has an equation to describe the relationship between flow, density, and speed, as
In route planning, traffic flow characteristics of route segments and lanes are both fundamental. With the assistance of data from C-V2X and loop detectors, the average traffic speed of section p is described as:
where is the value of CV market penetration ratio. The traffic speed
is calculated by data from loop detectors installed on the midstream of all lanes, while traffic speed
is calculated by data of CVs (except upstream and downstream) from C-V2X. In detail,
and
are calculated as:
For road-level traffic estimation, the number of vehicles stands for vehicles numbers detected of the road section, while for lane-level traffic estimation, the number of vehicles stands for vehicles numbers detected by the lane section. Based on traffic speed, we use the enhanced Van Aerde model [35] shown as follows to estimate the density of segments and lanes in the road network and to capture sudden congestion effects (e.g., accidents, lane closures) by integrating real-time incident data from road sensors. The revised density estimation is defined as:
where is dynamically updated based on incident reports and sensor inputs, and:
This adjustment allows the model to rapidly adapt to abrupt traffic state changes.
Route planning at road level
Since the state of traffic changes over time, dynamic route planning in the emergency response context is indispensable. Fig 3 shows the two steps in dynamic emergency route planning. At first, we plan route at road level, as Fig 3(a). The intersections are regarded as nodes in the road network. Specifically, according to the initial route, where paths and nodes are recorded, several following nodes are taken into consideration from the position of the EV to find the local optimal route. To simplify, in this paper, we take four nodes as an example at each sampling time for local road-level route planning. Note that, the position of the EV is denoted as origin and the fourth following node in the initial route from the origin is labeled as sub-destination. Then, all alternative routes that include less than seven nodes (maintain the same origin and the sub-destination) are provided for the EV to select. According to the traffic state estimation results of all alternative routes, the upper-layer MPC would output the travel time for the EV on each route. The route that costs the least time is selected as the suggested route segment and lane-level route planning will be executed. When the EV arrives at sub-destination, origin and sub-destination will be updated and the next stage of road-level route planning will be carried out. The above road-level dynamic route planning algorithm is summarized in Table 2.
Herein, the operation process of MPC is elaborated as follows. Assume that at time , there exist equations as:
where
There is a relationship between the density, flow and speed of two connected segments at time and that at time
according to traditional macroscopic traffic flow model [34,36]:
where
Note that, if there not exist ,
,
nor
, the value of them are set as the desired value of these parameters. Hence, the relationship can be expressed in the following way:
where
Using Eqs. (17) and (18), the state space representation of the traffic system can be obtained:
where
Set as predicted state variables outputs. Thus, Eqs. (17) and (18) can be formulated as:
Moreover, the objective function in the MPC is written as:
where
where represents lane-level queue lengths at intersection b, and
balances route efficiency and congestion avoidance. Therefore, the problem can be solved by Quadratic Programming [37].
After the suggested route segment is provided, we plan route at lane level as Fig 3(b). In this step, we combined lane selection with signal control which guarantees the EV’s efficiency without lane changing during driving. The details of the signal control scheme and lane-level route planning follow.
Route planning at lane level considering Signal preemption control
In normal, vehicles decelerate for lane changing which might cost more travel time. In this study, the optimal lane for the EV is suggested through the lane-level route planning algorithm. Moreover, to enhance the efficiency of the EV, a traffic light provides an applicable signal to guarantee that the EV could pass through multiple intersection without queueing and reduce deceleration performance. Although improving the efficiency of the EV is vital, the impact of EV operation on normal traffic is essential to be taken into consideration in signal preemption control. Therefore, we employ two signal control strategies combined with lane-level route planning to decrease the impact of EV operation on normal traffic as well as ensure the EV’s efficiency.
The process of lane-level route planning is utilized through lower-layer MPC which is similar to the upper-layer MPC except that the predicted state variables and past state variables belong to lane-level traffic flow. Further, each time after predicting arrival time through MPC, whether execute signal control and which control strategy is executed depends on the arrival phase and the value of that measures the overall benefits of the subjective intersection
after signal control. If the predicted arrival phase of the EV is green, there is no need for signal control; otherwise, signal control strategies will be considered. One of the two signal control strategies leveraged in this study is Green Extension, as Fig 4 (a), while the other is Red Truncation shown as Fig 4 (b). More specifically, the Green Extension strategy is carried out if the following three requirements are met:
- The arrival time of the last EV in EVs platoon is later than the ending time of its corresponding green phase.
- After Green Extension, the saturation flow of each phase is less than
.
- The value of
is greater than zero.
In this paper, we use as the performance index for normal traffic after traffic control of intersection
. Note that the value of
represents the weighted sum of delay for both normal vehicles and the EV at the intersection
:
where is the sum of the delay time at no-EV phase of intersection
, and
is the weighted sum of the reduced delay time at EV phase of intersection
. Thus, when value of
is greater than zero, it means that the adverse influence caused by traffic control is acceptable. The calculative process for
is elaborated as follow:
According to the Highway Capacity Manual 2016 (HCM2016) [26], the relationship between ,
, and
is described as:
Then the green time for extension can be calculated as:
Thus, the reduced delay time for the EV under Green Extension, which is equal to the time for the EV to queue without signal control, can be written as:
The above equation describes the waiting time in line when the EV does not enjoy preemption. In addition, after the Green Extension strategy is adopted, some normal vehicles in this direction also get more time for passing through the intersection without waiting for the next cycle. In this case, the illustration of cumulative arrival curve and departure curve are shown as Fig 5(a). According to these curves, the reduced delay time for normal vehicles after Green Extension on turn lane in phase
is formulated as:
Provided that the number of turn lanes in intersection is
and the number of no-EV phases in intersection
is
, then, the weighted sum of the reduced delay time of intersection
can be calculated as:
Note that the weight is determined by the urgency of the emergency rescue. Thus, delay time for normal vehicles on turn lane
intersection
at phase
and the sum of the delay time at no-EV phase of intersection b are obtained:
Hence, the value of under Green Extension strategy is calculated as:
Alternatively, the Red Truncation strategy is carried out if the following five requirements are met:
- The arrival time of the last EV in EVs platoon is earlier than the starting time of its corresponding red phase.
- All green time in no-EV phases is greater than the minimum green time
.
- The compressible red time is greater than zero.
- After Red Truncation, the saturation flow of each phase is less than 0.98 and the time for queue dissipation on the subjective lane
is less than the EV travel time
.
- The value of
is greater than zero.
In this case, the definition of is same as that under Green Extension strategy. More specifically, based on Eq. (35), the compressible red time can be calculated as:
Under Red Truncation strategy, normal vehicles in the same phase as the EV can also pass through the intersection in advance and the illustration of cumulative arrival curve and departure curve are shown as Fig 5(b). The shaded area represents the increased delay time for normal vehicles after Red Truncation on turn lane in phase
and the weighted sum of the increased delay time
are formulated as:
After Red Truncation, delay time for normal vehicles on turn lane intersection
at phase
and the sum of the delay time at no-EV phase of intersection b are shown as:
Then, the value of under Red Truncation strategy is calculated as:
Additionally, based on Wang’s research [38], the predicted time for queue dissipation on the subjective lane
is calculated as:
The lower-layer MPC not only optimizes EV travel time but also penalizes excessive queuing in general traffic lanes. The objective function includes:
where is the predicted queue length at intersection b, and
balances EV priority and traffic fairness.
The above lane-level dynamic route planning algorithm with signal preemption control is summarized in Table 3.
A study case to test the numerical model on a real network
This section examines the effectiveness of the proposed system based on the simulation software MATLAB. We consider five EVs in the same lane in a realistic scenario which is the region with 179 intersections in Kunshan, China. Herein, based on the origin and the destination, A* algorithm is employed to generate the initial route. Moreover, the initial signal timing and lanes number of each intersection are obtained from ground truth dataset. For simplicity, the value of desired speed, the value of desired density and the value of start-up lost time as well as the value of free flow speed for vehicles respectively are the same on all lanes in this road network during simulation. Assume that all vehicles follow vehicle kinematics equation and the distribution of vehicles on the road network at the initial time is Gaussian. The specific parameters and values are presented in Table 4.
In order to better demonstrate the validity of the proposed system, we perform numerical simulation experiments under different flow of traffic environments. The traffic flow conditions are set sequentially for 550 vehicles/(hour*lane), 1100 vehicles/(hour*lane) and 1600 vehicles/(hour*lane) to stand for low traffic case, medium traffic case and high traffic case, respectively according to the previous study [39,40]. Considering the randomness in signal time and vehicle arrival, each case is repeated multiple times and the results are averaged. Some indexes, i.e., delay time of each intersection for normal traffic and average delay time of the road network for normal traffic, in each case are demonstrated in Table V. Note, among the approaches in Table V, only A* means the EV travel on the initial route without lane suggestion and signal control; only signal preemption with A* means providing signal preemption in the initial route calculated by A* for the EV taking no account of passing through lanes for other directions at each intersection; lane-level route planning without signal control means providing lane-level route planning only according to the minimal EV’s traveling time in each road section; only signal preemption with road-level route planning means providing the EV with signal preemption under dynamic road-level route planning but dismissing passing through lanes for other directions at each intersection. Specifically, the average delay time of each intersection for normal traffic is described as:
To analyze the impact of signal preemption on general traffic, two new performance metrics: General Traffic Delay Change Rate (GTDCR) and Intersection Queue Length (IQL), are introduced in this study. The value of GTDCR can be expressed as Eq. 55 and IQL represents the maximum queue length during preemption cycles.
As shown in Table 5, compared with other approaches, the proposed system has a slight decrease in the average delay time of each intersection for normal traffic, and the delay time of the road network for normal traffic in Low Traffic and Medium Traffic cases. compares the impact of different approaches on general traffic. The proposed system achieves the lowest GTDCR (7.32% to 12.54%) and IQL (14.25–21.73 vehicles) across all traffic conditions, demonstrating its ability to balance EV priority with minimal disruption. In contrast, static methods (e.g., Only signal preemption with A*) exacerbate congestion, with GTDCR exceeding 33.80% and IQL reaching 42.66 vehicles in high traffic. It is worth noting that there is a substantial decline in the two indexes using the proposed system compared with using the other three approaches, since the proposed system could offer lane-level route selection and reduce congestions of road networks, in the High Traffic case. The EVs could maintain efficiency without queuing in all traffic flow cases and cause less delay time for normal traffic when using the proposed system. Further, under higher traffic flow conditions, the times that a signal control strategy is exercised (i.e., 2.9, 7.3, 14.1 in Low Traffic, Medium Traffic, and High Traffic cases, respectively) would be more and the delay time of each intersection and the whole road network are both longer than under lower traffic flow condition, since signal control strategy could influence the behaviors of normal vehicles.
Fig 6 shows the average travel time and the average velocity of EV to compare the efficiency of EVs under different traffic flow conditions. Herein, results of lane-level route planning without signal control is not presented since it could bring less travel time to EV but more delay time for normal traffic. As shown in Fig 6, as the traffic flow increases, the average velocity of EVs will decrease and the travel time will increase. However, in all cases, EVs using the proposed system performs better than others. Particularly, the average velocity of EVs is close to the desired speed in low traffic case using the proposed system.
Discussions
Compatibility with Partial CV Infrastructure
While CV technology enhances real-time coordination, the proposed system prioritizes compatibility with existing urban sensing infrastructure. For example:
- Loop Detectors: Provide foundational traffic flow data for road-level route planning.
- Mobile Signaling: Compensates for sparse CV data by estimating lane-level vehicle trajectories (Section 3.1).
- Fallback Mechanism: If CV communication is unavailable (e.g., due to device failures), the system dynamically shifts to historical traffic patterns and loop detector inputs.
To further verify the effectiveness of the proposed system, five CV market penetration ratios under high traffic cases are examined: 10%, 30%, 50%, 70%, 90%. Likewise, the performance of our proposed system is compared with that of the EV control strategy proposed by [39] labeled as AEVCS, as shown in Fig 7. It can be seen that under high traffic cases, the EVs using our proposed system reach their destination faster and cause less delay time for normal traffic. Moreover, when there exist more CVs, the traffic state estimation results would be more accurate, so that the EVs efficiency can be improved and the adverse impact on normal traffic is reduced.
Compatibility with communication latency and multi-EVs
To evaluate the system’s robustness under realistic communication conditions, we conducted additional experiments introducing 0.2-second latency time respectively in V2I/V2V data transmission as shown in Fig 8. This latency simulates typical delays caused by network congestion or signal processing in 90% CV penetration ratio environments with multi-EVs.
The simulation results indicate that a communication latency of 0.2 seconds introduces moderate performance degradation (8.3% increase in EV travel time and 7.3% rise in general traffic delays). However, these impacts remain within acceptable thresholds, as the system retains its ability to prioritize EVs while mitigating congestion compared to static strategies. Regarding scalability, increasing the number of EVs amplifies their collective impact on normal traffic. Under low-to-medium traffic conditions (e.g., 550–1100 vehicles/hour/lane), the system effectively minimizes disruptions. In high-density scenarios (1600 vehicles/hour/lane), the cumulative effect of multiple EVs raises intersection delays by 6.7% compared to single-EV cases. Nevertheless, the proposed bi-level MPC framework dynamically redistributes traffic pressure through lane-level rerouting and adaptive signal preemption, ensuring that adverse impacts remain manageable. This balance underscores the system’s robustness in heterogeneous environments.
Suppose that the EV platoon is made up of six EVs and the CV market penetration ratios is 50%. As shown in Fig 7, comparing the proposed system with only signal preemption with road-level route planning when the traffic flow is not high, the changes in EV efficiency are not obvious; but in high traffic cases, EV efficiency has been significantly improved. This phenomenon may be due to when providing lane-level route planning, there are fewer vehicles in front of EVs and the road environment is relatively simple. Hence, there is less interference to the speed of EVs in high traffic cases. Then, compared with only signal preemption with A*, EV efficiency is further reduced, owing to dynamic traffic information update and planning, particularly in high traffic cases. Further, without dynamic route planning and signal preemption control, EVs are the least efficient, since they may experience parking and queueing.
In brief, the integrated system of this paper has some advantages. First, it applies rolling horizon to route planning, which can help to adjust the planned route dynamically and improve the efficiency of EVs. Second, the adverse impact of signal preemption on normal traffic is taken into consideration. Through the changes in parameter (i.e., number of nodes for road-level route planning and sampling time) settings, the different situations can be simulated and analyzed. In this case, we examine the case when the number of nodes for road-level route planning and sampling time change.
With the same sampling time, the local road-level planned route might be affected by the number of selected nodes. Fig 9 shows the average travel time and the average velocity for the EVs under different traffic flow conditions considering the number of nodes selected for road-level route planning, with the sampling time of 15s. If fewer nodes are taken into consideration, there is fewer alternative routes for EV to select which has limitations to the choice of road for optimal EV efficiency. Therefore, it takes less traveling time with more selected nodes for EV and the average velocity is promoted, since more route information is given. However, too much route information may slow down the speed of calculation and solution. The number of selected nodes should be set according to the real road environment.
With the same number of nodes selected, the EVs efficiency could also be affected by sampling time for MPC. Fig 10 shows the average travel time and the average velocity for the EVs under different traffic flow conditions considering sampling time, with four nodes selected. Less sampling time stands for more exquisite perception and optimization. The simulation results show that EV efficiency will increase with less sampling time. However, as a matter of fact, advising an EV on speed and trajectory constantly might bring forced communication abnormalities.
Dynamic route planning and signal preemption control are believed to play a significant role in EVs. On the basis of the simulated results, the integrated system of signal preemption control and route planning can help to maintain EV efficiency and reduce the impact of EVs operation on normal traffic. However, many problems remain to be addressed before wide-scale application is possible—such as the complexity of road network topology building and the high cost of CV constructing. At present, the traffic information can only be constructed using expert data, which is usually acquired by means of multi-layered LiDAR and is very difficult to update online. In future, CV technology will be employed to extract useful information and transfer information.
Limitations of communication reliability assumptions
The dynamic congestion adjustment enables the system to better handle real-world disruptions. For example, in a simulated accident scenario, the revised model reduced rerouting delays by 15% compared to the original framework. Future work could integrate crowd-sourced incident reports (e.g., via mobile apps) to further improve traffic flow estimation accuracy.
The proposed system assumes reliable communication (i.e., no delays or packet loss) between CVs and infrastructure, with no delays or packet loss. However, in real-world scenarios, communication reliability in partial CV environments may be compromised due to factors such as signal interference, hardware failures, or network congestion. This limitation could affect the accuracy of traffic state estimation and the effectiveness of signal preemption control. For instance, intermittent data loss might lead to outdated traffic flow predictions, delaying dynamic route updates for EVs. Recent studies (e.g., [41] Finkelberg et al., 2022; [42] Petrov et al., 2024) highlight the critical impact of communication reliability on traffic control systems. Finkelberg et al. (2022) demonstrated that even a 10% packet loss rate could degrade intersection control performance by 15%. To address this, future work should integrate robust communication models (e.g., stochastic MPC) or adaptive fault-tolerant mechanisms. Potential solutions include:
- Dynamic Data Fusion: Combining real-time CV data with historical patterns to mitigate sparse data issues.
- Fallback Protocols: Triggering pre-defined emergency routes if communication failures exceed a threshold.
- Edge Computing: Deploying localized decision-making at roadside units to reduce dependency on central servers.
These enhancements would improve the system’s resilience in heterogeneous communication environments.
Conclusions
In this study, we have proposed an integrated system of signal preemption control and route planning to address the emergency route planning problem for an EV platoon using CV technology in emergency rescue under partial CV environment. A bi-layer MPC optimization framework that dynamically predicts the optimal road and lane for EVs respectively has been developed by traffic flow estimation via data from C-V2X and loop detectors. Further, the signal preemption control scheme combined with lane selecting has been designed to seek an approach to improve the efficiency of EVs and to reduce the impact of EVs operation on normal traffic. Finally, a numerical example and simulations have been provided to illustrate the effectiveness and improvements of the proposed system under different CV market penetration ratios. Potential future work can be focused on (i) how to better utilize vehicle information under C-V2X environment to promote the estimation accuracy of traffic flow characteristics in the road network; (ii) how to take into consideration heterogeneous traffic that consists of automatic vehicles, CVs and traditional human-driven vehicles in the design; (iii) how to better control signal and plan route for EVs if the signal timing of each intersection in the road network is not fixed but changes according to the traffic condition.
References
- 1. Abudayyeh D, Ngoduy D, Nicholson A. Traffic Signal Optimisation in Disrupted Networks with Re-Routing. Transportation Research Procedia. 2018;34:195–202.
- 2. Norii T, Nakao S, Miyoshi T, Braude D, Sklar DP, Crandall C. Driving Ambulances Safely: Findings of Ten Years of Japanese Ambulance Crash Data. Prehosp Emerg Care. 2023;27(1):94–100. pmid:34874807
- 3. Gai W, Deng Y. Survey-based analysis on the diffusion of evacuation advisory warnings during regional evacuations for accidents that release toxic vapors: A case study. Journal of Loss Prevention in the Process Industries. 2019;57:174–85.
- 4. Li M, Wu X, He X, Yu G, Wang Y. An eco-driving system for electric vehicles with signal control under V2X environment. Transportation Research Part C: Emerging Technologies. 2018;93:335–50.
- 5. Jose C, Vijula Grace KS. Optimization Based Routing Model for the Dynamic Path Planning of Emergency Vehicles. Evol Intell. 2022;15(0):1425–39.
- 6. Mu H, Song Y, Liu L. Route-Based Signal Preemption Control of Emergency Vehicle. Journal of Control Science and Engineering. 2018;2018:1–11.
- 7. Wu J, Liu P, Zhou Y, Yu H. Stationary condition based performance analysis of the contraflow left-turn lane design considering the influence of the upstream intersection. Transportation Research Part C: Emerging Technologies. 2021;122:102919.
- 8. Næss LE, Krüger AJ, Uleberg O, Haugland H, Dale J, Wattø J-O, et al. Using machine learning to assess the extent of busy ambulances and its impact on ambulance response times: A retrospective observational study. PLoS One. 2024;19(1):e0296308. pmid:38181019
- 9.
Mahariba A, Uthra R, Brunet R. Estimation of shortest route with minimum travel time using GIS and MSSTT algorithm. In: Loon L, Subramaniyan M, Gunasekaran K, editors. Advances in construction management. Singapore: Springer Nature. 2022. p. 565–79.
- 10. Xu K, Gai W, Salhi S. Dynamic emergency route planning for major chemical accidents: Models and application. Safety Science. 2021;135:105113.
- 11. Zhao J, Guo Y, Duan X. Dynamic Path Planning of Emergency Vehicles Based on Travel Time Prediction. Journal of Advanced Transportation. 2017;2017:1–14.
- 12.
Fontanelli S, Bini E, Santi P. Dynamic route planning in vehicular networks based on future travel estimation. In: 2010 IEEE Vehicular Networking Conference. IEEE. 2010. 126–33. https://doi.org/10.1109/vnc.2010.5698247
- 13.
Lakshmi A, Sekhar P, Joseph K, Priya A. A systematic review of route optimization for ambulance routing problem. In: Atlantis Press. 2023. 294–304.
- 14. Chen T, Cai Y, Chen L, Xu X. Trajectory and Velocity Planning Method of Emergency Rescue Vehicle Based on Segmented Three-Dimensional Quartic Bezier Curve. IEEE Trans Intell Transport Syst. 2023;24(3):3461–75.
- 15.
Salehinejad H, Pouladi F, Talebi S. Intelligent Navigation of Emergency Vehicles. In: 2011 Developments in E-systems Engineering. IEEE. 2011. 318–23. https://doi.org/10.1109/dese.2011.17
- 16. Liu Y, Zhang K, Hou B, Li Q, Feng J, Nguyen T-M-T, et al. Real-time traffic impedance and priority based cooperative path planning mechanism for SOC-ITS: Efficiency and equilibrium. Simulation Modelling Practice and Theory. 2023;122:102683.
- 17. Jiang K, Yang D, Liu C, Zhang T, Xiao Z. A Flexible Multi-Layer Map Model Designed for Lane-Level Route Planning in Autonomous Vehicles. Engineering. 2019;5(2):305–18.
- 18. Olivier A, Mohammadi S, Smyth AW, Adams M. Bayesian neural networks with physics‐aware regularization for probabilistic travel time modeling. Computer aided Civil Eng. 2023;38(18):2614–31.
- 19. Małecki K, Kamiński M, Wąs J. A multi-cell Cellular Automata model of traffic flow with emergency vehicles: Effect of a corridor of life and drivers’ behaviour. Journal of Computational Science. 2022;61:101628.
- 20. Liang X (Joyce), Guler SI, Gayah VV. A heuristic method to optimize generic signal phasing and timing plans at signalized intersections using Connected Vehicle technology. Transportation Research Part C: Emerging Technologies. 2020;111:156–70.
- 21. Chen Y, Wu G, Chen Y, Xia Z. Spatial Location Optimization of Fire Stations with Traffic Status and Urban Functional Areas. Appl Spatial Analysis. 2023;16(2):771–88.
- 22. Shaaban K, Khan MA, Hamila R, Ghanim M. A Strategy for Emergency Vehicle Preemption and Route Selection. Arab J Sci Eng. 2019;44(10):8905–13.
- 23.
Heda M, Busch SJ, Sepa Seid Mayo A, Ermakova T. Rescue route: using real-time traffic data for routing emergency vehicles. In: 2023 IEEE/WIC International Conference on Web Intelligence and Intelligent Agent Technology. IEEE. 2023. 362–6.
- 24. Su H, Zhong YD, Chow JYJ, Dey B, Jin L. EMVLight: A multi-agent reinforcement learning framework for an emergency vehicle decentralized routing and traffic signal control system. Transportation Research Part C: Emerging Technologies. 2023;146:103955.
- 25. Zhan FB, Noon CE. Shortest Path Algorithms: An Evaluation Using Real Road Networks. Transportation Science. 1998;32(1):65–73.
- 26. Mirzahossein H, Shahrad S, Afandizadeh Zargari Z, Motevalli Habibi H. Determination of passenger car equivalent for heavy vehicles in iran in comparison with highway capacity manual 2016. Iranian J Sci Technol, Trans Civ Eng. 2016;:1–11.
- 27. Dutta P, Khatua S, Choudhury S. Fast move: A prioritized vehicle rerouting strategy in smart city. Vehicular Communications. 2023;44:100666.
- 28. Qi W, Landfeldt B, Song Q, Guo L, Jamalipour A. Traffic Differentiated Clustering Routing in DSRC and C-V2X Hybrid Vehicular Networks. IEEE Trans Veh Technol. 2020;69(7):7723–34.
- 29. Lin P, Chen Z, Pei M, Ding Y, Qu X, Zhong L. Multiple Emergency Vehicle Priority in a Connected Vehicle Environment: A Cooperative Method. IEEE Trans Intell Transport Syst. 2024;25(1):173–88.
- 30. Cortés CE, Stefoni B. Trajectory Simulation of Emergency Vehicles and Interactions with Surrounding Traffic. Journal of Advanced Transportation. 2023;2023:1–23.
- 31. Xie Z, Jin H, Teng J, Yang X. Design and Management of Multi-functional Exclusive Lane for the Integrated Service to Various Vehicles with Priority. KSCE Journal of Civil Engineering. 2022;26(2):882–92.
- 32. Humagain S, Sinha R, Lai E, Ranjitkar P. A systematic review of route optimisation and pre-emption methods for emergency vehicles. Transport Reviews. 2019;40(1):35–53.
- 33. Zeng Z, Yi W, Wang S, Qu X. Emergency Vehicle Routing in Urban Road Networks with Multistakeholder Cooperation. J Transp Eng, Part A: Systems. 2021;147(10).
- 34. Hara Y, Suzuki J, Kuwahara M. Network-wide traffic state estimation using a mixture Gaussian graphical model and graphical lasso. Transportation Research Part C: Emerging Technologies. 2018;86:622–38.
- 35.
Van Aerde M, Rakha H. Multivariate calibration of single regime speed-flow-density relationships [road traffic management]. In: Pacific Rim TransTech Conference. 1995 Vehicle Navigation and Information Systems Conference Proceedings. 6th International VNIS. A Ride into the Future. IEEE. 334–41. https://doi.org/10.1109/vnis.1995.518858
- 36. Puppo G, Semplice M, Tosin A, Visconti G. Fundamental diagrams in traffic flow: The case of heterogeneous kinetic models. Communications in Mathematical Sciences. 2016;14(3):643–69.
- 37.
Razzaghpour M, Shahram S, Valiente R, Fallah YP. Impact of communication loss on MPC based cooperative adaptive cruise control and platooning. In: arXiv.
- 38.
Wang S, Xu J, Yang G, Chen C. A green wave band model considering variable queue clearance time. In: Proceedings of the 10th World Congress on Intelligent Control and Automation. IEEE. 2012. 3025–30. https://doi.org/10.1109/wcica.2012.6358390
- 39. Zhang Y, Su R. An optimization model and traffic light control scheme for heterogeneous traffic systems. Transportation Research Part C: Emerging Technologies. 2021;124:102911.
- 40. So J, Kang J, Park S, Park I, Lee J. Automated emergency vehicle control strategy based on automated driving controls. J Adv Transp. 2020;2020(0):1–11.
- 41. Finkelberg I, Petrov T, Gal-Tzur A, Zarkhin N, Pocta P, Kovacikova T, et al. The Effects of Vehicle-to-Infrastructure Communication Reliability on Performance of Signalized Intersection Traffic Control. IEEE Trans Intell Transport Syst. 2022;23(9):15450–61.
- 42. Petrov T, Finkelberg I, Počta P, Buzna Ľ, Zarkhin N, Gal-Tzur A, et al. An analytical approach to the estimation of vehicular communication reliability for intersection control applications. Vehicular Communications. 2024;45:100693.