An efficient AI model for highway vehicle speed limit violation detection and real-time law-enforcing functionality

Used approaches	Discussion
A unified digital framework for traffic conflict prediction using a deep learning approach [17].	This technique has promise to improve traffic management and road safety, despite constraints such as data heterogeneity, real-time processing, and model interpretation difficulty.
Automatic vehicle's number plate recognition using artificial neural network and fuzzy approaches. [4, 5].	These methodologies are effective for real-time traffic management. Efficacy of number plate scanning is entirely contingent upon the illumination and clarity of the plate and the effectiveness of the used scanning method.
Deep learning approaches, used for the network traffic surveillance and analysis [18, 19].	These methods are effective for monitoring and analyzing traffic. Certain applications for monitoring and analyzing network traffic encounter low efficacy in employing deep learning due to insufficient labeled data, resource-constrained networks, retraining challenges, and others.
Intelligent transportation systems enabled by the IoT: Real-time traffic surveillance and regulation management [20,21,22]	Despite their efficacy in real-time traffic management, current methods are not environmentally friendly and fail to account for different road conditions. Considering the constraints of the testing equipment and environmental factors, it is necessary to assess alternative scenarios and enhancements to the present testing method to validate the system.
Edge machine learning methodology for intelligent transportation systems in smart traffic management [19, 23]	The primary obstacle with this strategy is the lighting conditions. This advanced model is significantly hindered by inadequate lighting during data acquisition. Moreover, connected vehicles can deliver real-time traffic conditions, enabling traffic lights to modify their timing in accordance with fluctuating traffic patterns.

[i] IoT, Internet of Things.

Central vehicle speed monitoring system [16].

Table 2:

Comparison of various traffic monitoring approaches for the highway

Publication details		Used approaches							Type of data set		Monitoring type			Offered	Prevent
Year	Ref. No.	Deep and/machine learning	Cloud/Fog/network storage	Image processing/pattern matching	Fuzzylogic	Block chain	Statistical method	Vehicular Ad-hoc network	Synthetic	Telematic	Real time	Vehicle over speed	Traffic rule violation	data security	data loss
2017	[13]
	[15]
2018	[33]
2019	[21]
	[25]
2020	[1]
	[3]
	[9]
	[12]
	[17]
	[39]
	[45]
2021	[2]
	[26]
	[38]
2022	[11]
	[14]
	[37]
	[46]
	[47]
2023	[30]
	[49]
2024	[7]
	[8]
	[19]
	[28]
	[32]
	[41]
2025	[5]
	[20]
	[34]
	[50]
	Proposed

[i] Herre graphic/j_ijssis-2026-0076_ingr_001.png denotes true, whereas denotes the partially true.

Table 3:

A summary of various challenges of cloud resource provisioning and their mitigation

Publication Year/reference		Monitoring vehicle speed smartly and accurately	Worn the driver for violating the speed limit	Imposing a penalty for continuous speed limit violation	Cost efficiency of the used technique
2017	[13, 15]	○		○
2018	[33]		○	○	○
2019	[21, 25]			○
2020	[1, 3, 9, 12, 17, 39, 45]				○
2021	[2, 26, 38]		○	○
2022	[11, 14, 37, 46, 47]		○	○
2023–24	[7, 8, 9, 28, 30, 32, 41, 49]		○	○
2025	[5, 20, 34, 50]		○	○
Our Work (2025)

Table 4:

Comparison of various existing approaches for controlling overspeed vehicles

Related topics	Strength	Shortcomings
Forecasting the traffic issues in real-time [3, 9, 12, 17]	A R-CNN model can forecast traffic conflicts by analyzing images captured by a single front-facing camera and identifying lane markings. A collection of deep neural network models for predicting traffic conflicts is created from this data by integrating it with traffic characteristics and calculated SSMs.	Unfortunately, the existing methodology does not provide a means to prevent data loss during transmission. This technology yields low precision and is not especially effective for the real-time transmission of traffic data.
Real-time traffic tracking and management with IoT for adaptive transit systems [5, 25, 31].	The objective of a strategic traffic management plan that integrates autonomous driving with the IoT is to alleviate traffic congestion. Traffic flow dynamics results with an appropriate control strategy for vehicle-to-everything-enabled cars from a macro viewpoint.	Challenges regarding the management of Intelligent Transportation Systems and the limited data transmission capabilities of IoT devices continue to exist. The incorporation of data transport reliability must be integrated into this model additionally.
The IoT-based wireless sensor network framework for monitoring the traffic surveillance information [13, 16, 17].	This algorithm analyzes test data to extract pertinent information, as significant signal noise during accelerometer monitoring complicates signal recognition. It further employs a peak detector and a moving average smoothing filter to precisely identify each passing axle and calculate the time interval between two axles.	Due to limitations of the testing apparatus and ambient variables, improvements to the current methodology are needed. The absence of a mechanism to guarantee data integrity in this technique adversely impacts the accuracy and quality.
Optimizing the traffic management in today's transportation networks with innovative machine learning [5, 12, 14].	The traffic signals are modified based on traffic calculations to reduce overall delay at the intersection. The traffic lights may interact with adjacent signals to convey the total duration of delays. This delay facilitates the priority of traffic in essential areas such as schools, offices, and hospitals.	Traffic trends can be forecasted via machine learning algorithms that examine both historical and current data. Connected vehicles provide real-time data on the traffic situation, enabling traffic signals to evolve patterns.
A system of interlinked cars utilized for the intelligent transportation system for the urban areas [15, 16, 27].	It presents an intelligent vehicular network for smart cities that selects routes via ad-hoc, real-time data, obtained from nearby vehicles. It utilizes Android-powered smartphones equipped with Wi-Fi direct functionality as in-vehicle devices.	This study does not aim to tackle communication issues, including delay, channel noise, and transmission failures. A possible outcome is a reduction in output quality, while another is data corruption.
Intelligence-based traffic management systems for signalized crossroads including non-autonomous vehicles [6, 12, 16].	This research presents a neural network capable of predicting delays for human-operated cars at signalized intersections. An ANN will address complex tasks, including the processing of extensive datasets with variable inputs.	It disregards the numerous issues that arise when transferring substantial volumes of data across locations. The concerns may adversely affect both data quality and the accuracy of the proposed model.
Utilization of IoT and multi-agent reinforcement learning, management of traffic lights might become automated [14, 32, 33].	This work presents a reinforcement learning methodology for the deployment and control of traffic signals employing numerous agents. The last phase involves developing an innovative reward function applicable in multiagent learning to effectively determine the optimal strategy for managing traffic light behaviors.	It utilizes roadside gadgets and vehicle-to-everything connectivity to improve data for traffic scenario model learning. Before implementing traffic signal control, the transportation system must be validated using a probabilistic model. The accuracy and dependability of the method can be improved.

[i] ANN, artificial neural network; IoT, Internet of Things; R-CNN, region-based convolutional neural network; SSM, surrogate safety measures.

The proposed model. ANN, artificial neural networks; IoT, Internet of Things.

**(A)** Bluetooth device, installed at the traffic light, **(B)** Bluetooth device, installed at the toll booth entrance, **(C)** Bluetooth device, installed at the banner stand, and **(D)** Bluetooth device, installed at the road divider.

The anticipated ANN model. ANN, neural network.

The integrated module, installed inside the car.

Algorithm-1:

Simulated Annealing for Vehicle Overspeed Detection

Input Data: →Number of speed limit violation, epochs;

Result:→ Generating Summon Accurately after violating the speed limit contagiously 3 times;

1. Let S ←S₀ be the initial solution where the driver violates the speed limit for the first time;

2. T←epochs;

3. While T> 0 do

a. Calculate speed limit value from the road sensor and set it as the maximum speed limit (m);

b. While number of violation (n) <3 do

(i) Select any other recorded speed (x_j) at the random basis in the next iteration;

(ii) If x_j > m

If the root is congested enough

Calculate the penalty value (P_j);

End if

If the root is not congested enough

Calculate the penalty value (P_k);

End if

Else if

e \frac{P_{j} - P_{k}}{k . T .} > Random (0, 1)

, where k is any random integer

Calculate the penalty value (P_f);

Penalty ← P_f.

Break;

End if

T ←T-1;

End While

Algorithm-2:

Error Control Bits Generation and Incorporation

Input: →Contiguous two 8-bit string and denoted them Str₁ and Str₂;

Output: →The complete challan string with error control bits as FinErr

(1) Initialize 〈FinErr〉 = ∅ and a string variable temp = ∅;

(2) for (int m = 0;

m < \bar{{Str}_{1}}

; m++) // concatenate both string

FinErr[m] = Str₁[m]

FinErr[Str₁ + m] = Str₂;

End for

(3) for (int i = 0;

i < \bar{{Str}_{1}}

; i++) //Performing XOR operation between both strings

temp[i] = Str₁[i]⊕Str₁[i];

End for

(4) for (int j = 0;

j < \bar{temp}

; j++) // Concatenate temp and the FinErr

FinErr \bar{{Str}_{1}} + \bar{{Str}_{2}} + j] = temp [j];

;

End for

(5) Repeat Step-1 to Step-4 all rest of the strings;

(6) Concatenate all the strings in final single string FinErr;

Table 5:

Cost estimation of the hypothetical architecture

Many-to-one Bluetooth device (for roadside)	Integrated module (for car)
Device cost: Rs. 2,000–3,000 (Max) Lithium battery (Life 2–3 years): Rs. 600 (Max) Installation and assemble cost: Rs. 2,000 Maintenance cost: Rs. 0	Uno Minda 7687Q-028-11-M01 SPEED SENSOR For-BAJAJ-DISCOVER 125ST-Rs. 300 (Max) LCD display unit (2.4 in.): Rs. 1,000 (Max) Storage unit (8 GB): Rs. 1,000 (Max) One-to-many Bluetooth device: Rs. 800 (Max) LM339 - Low power Quad processor: Rs. 100 (Max) Lithium battery (Life 2–3 years): Rs. 600 (Max) Installation and assemble cost: Rs. 2,000 Maintenance cost: Rs. 0
Total cost: Rs. 5,600 (Max)	Total cost: Rs. 6,700 (Max)

Accuracy percentage, provided by several AI methods for overspeed detection. AI, artificial intelligence; CNN, condensed nearest neighbour rule; COID, case method based on clustering and outliers and internal detection; IBL, instance-based learning; R-CNN, region-based convolutional neural network; RNN, recurrent neural network; WCOID, weighting and clustering outliers and internal cases detection.

Table 6:

Accuracy percentage offered by distinct AI approaches

Used AI techniques	Accuracy percentage
WCOID	90.7
COID	90.6
CNN	72.5
RNN	74.2
R-CNN	87.1
IBL	79.3
Proposed combo	99.2

[i] AI, artificial intelligence; COID, case method based on clustering and outliers and internal detection; IBL, instance-based learning; RNN, recurrent neural network; R-CNN, region-based convolutional neural network; WCOID, weighting and clustering outliers and internal cases detection.

Comparison of different AI approaches in terms of precision, recall, F1-score, and AUC-score. AI, artificial intelligence; AUC, area under the ROC curve; CNN, condensed nearest neighbour rule; COID, case method based on clustering and outliers and internal detection; IBL, instance-based learning; R-CNN, region-based convolutional neural network; RNN, recurrent neural network; WCOID, weighting and clustering outliers and internal cases detection.

Table 7:

Comparison of AI techniques in terms of their capacity to produce precision, recall, F1-Score, and AUC-Score

Used techniques	Precision	Recall	F1-score	AUC-score
WCOID	0.89	0.91	0.89	0.87
COID	0.85	0.88	0.87	0.85
R-CNN	0.83	0.85	0.86	0.83
IBL	0.81	0.83	0.85	0.81
RNN	0.78	0.79	0.79	0.78
CNN	0.76	0.78	0.78	0.76
Proposed technique	0.98	0.99	0.99	0.99

[i] AI, artificial intelligence; AUC, area under the ROC curve; CNN, condensed nearest neighbour rule; COID, case method based on clustering and outliers and internal detection; IBL, instance-based learning; R-CNN, region-based convolutional neural network; RNN, recurrent neural network; WCOID, weighting and clustering outliers and internal cases detection.

Percentage of data loss provided by different AI-based methods during data transmission. AI, artificial intelligence; CNN, condensed nearest neighbour rule; COID, case method based on clustering and outliers and internal detection; IBL, instance-based learning; R-CNN, region-based convolutional neural network; RNN, recurrent neural network; WCOID, weighting and clustering outliers and internal cases detection.

Table 8:

Comparison of AI techniques with their data loss percentage

AI techniques	% of data loss
WCOID	0.0075
COID	0.007
CNN	0.005
RNN	0.006
R-CNN	0.0045
IBL	0.008
Proposed combo	0.0006

[i] AI, artificial intelligence; CNN, condensed nearest neighbour rule; COID, case method based on clustering and outliers and internal detection; IBL, instance-based learning; R-CNN, region-based convolutional neural network; RNN, recurrent neural network; WCOID, weighting and clustering outliers and internal cases detection.

SNR_dB, offered by various error control techniques. SNR, signal-to-noise ratio.

Table 9:

Comparison of distinct error control techniques along with their SNR (DB)

Used techniques	SNR (DB)
Parity checking	36.1
Low density parity checking	38.9
Cyclic redundancy	39.1
Forward error correction	44.1
Hamming distance	52.2
Reed-Solomon	53.4
Proposed technique	61.95

[i] DB, Decibel; SNR, signal-to-noise ratio.

Percentage of packet loss offered by various data retrieval schemes. FEC, forward error correction; LDPC, low-density parity checking.

Table 10:

Comparison of distinct error control techniques along with their packet loss percentages

Used techniques	Percentage of packet loss
Low density parity checking	0.0064
Reed-Solomon	0.0043
Forward error correction	0.0036
Hamming distance	0.0031
Cyclic redundancy check	0.0024
Proposed technique	0.00091

Table 11:

CC table for different AI approaches

Applied AI techniques	CC
WCOID	5
COID	6
CNN	4
RNN	4
R-CNN	4
IBL	7
Proposed combo	2

[i] AI, artificial intelligence; CC, cyclomatic complexity; CNN, condensed nearest neighbour rule; COID, case method based on clustering and outliers and internal detection; IBL, instance-based learning; R-CNN, region-based convolutional neural network; RNN, recurrent neural network; WCOID, weighting and clustering outliers and internal cases detection.

Information processing capacity of different AI approaches. AI, artificial intelligence.

Table 12:

Comparison of various AI techniques along with their capacity for information processing

AI techniques	Processing capacity (Info/second)
WCOID	33
COID	29
CNN	39
RNN	37
R-CNN	41.2
IBL	20
Proposed combo	57

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/ijssis-2026-0076 | Journal eISSN: 1178-5608

Journal RSS Feed

Language: English

Submitted on: Sep 12, 2024

Published on: Sep 22, 2026

Published by: International Journal on Smart Sensing and Intelligent Systems

In partnership with: Paradigm Publishing Services

Publication frequency: Volume open

Keywords:

speed limit violators,

psychological impact on violators,

IoT-based real-time responsive AI approach,

speed limit violation detection accuracy,

data integrity,

simulated annealing

Related subjects:

Engineering,

Introductions and overviews,

Engineering, other

© 2026 Shiladitya Bhattacharjee, Sulabh Bansal, Tanupriya Choudhury, Ketan Kotecha, Piyush Chauhan, published by International Journal on Smart Sensing and Intelligent Systems
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.

Volume 19 (2026): Issue 1 (January 2026)

An efficient AI model for highway vehicle speed limit violation detection and real-time law-enforcing functionality

Figures & Tables

Figure 1:

Table 1:

Figure 2:

Table 2:

Table 3:

Table 4:

Figure 3:

Figure 4:

Figure 5:

Figure 6:

Algorithm-1:

Algorithm-2:

Table 5:

Figure 7:

Table 6:

Figure 8:

Table 7:

Figure 9:

Table 8:

Figure 10:

Table 9:

Figure 11:

Table 10:

Table 11:

Figure 12:

Table 12:

Paradigm

My account