One transition from gas-powered vehicles to EVs is a clear example of a transformational activity in transportation on the world stage. This change is guided by decarbonization and energy efficiency needs. In light of the above, the IoEV concept has arisen as a disruptive one. Electrically powered vehicles with communication, cloud intelligence, and data-driven optimization embedded in the network represent emblazoned concepts. From its perspective, the IoEV is not only about electric vehicles but about a digitally connected ecosystem where vehicles, charging stations, grid systems, and users interact jointly and in real-time so that energy and mobility decisions can be made [1,2].

The evolution of IoEVs is tightly intertwined with anything that concerns smart cities, which in turn demand intelligent transportation infrastructures that are sustainable, autonomous, and interconnected. In such a city, electric vehicles remain in possession of consumption power and hold a role in aiding with respect to grid balancing through V2G interaction. Moreover, the widespread penetration of IoT sensors and edge computing allows for remote monitoring of EV performance, battery status, and real-time charging control [2].

Domain three core concerns characterize the performance landscape of IoEVs: energy efficiency, battery longevity, and grid integration stability. First, the energy consumption of an EV depends on user behavior, traffic conditions, and battery state, thus requiring dynamic energy management [3]. Second, battery health impacts the operating life of an EV and, in turn, its resale value; charging cycles with low degradation, heavy current loads, and thermal stress can accelerate battery aging. Finally, inconsistent and sudden demand spikes from many EVs connecting to the grid can become load imbalances, voltage drops, and costing issues for utilities [4].

In the face of such problems, two solutions arose to become the key enablers for the sustainable IoEV deployment: smart charging infrastructure and intelligent BMS. A smart charging station controls charging rates dynamically with respect to the grid conditions, electricity prices, and vehicle needs. Charging can occur when the grid is free; vehicles can be qualified according to emergency standing, or even renewable energy usage can be supported [5,6].

Incorporating these two solutions into a communication-aware IoEV solution framework leads to better energy conservation in the whole EV ecosystem, lesser battery ageing, and supporting symmetry constraints on the grid side. See how Fig. 1 depicts a conceptual framework showing how an electric vehicle interacts with a smart charging station, grid system, and cloud server [7]. An onboard EV BMS communicates the SOC/SOH to the charging station. The charging station schedules the charging operations from the time-varying grid data and pricing data, considering vehicle priorities and coordinates all this via IoT/cloud communication [8].

The existing EV charging infrastructures are predominantly based on the fixed-rate or user-initiated charging that do not take into account either the battery’s immediate situation or the distribution grid’s state. Thus, the charging events very often bring about high C-rates, thermal stress, and unsynchronized consumptions that lead to battery depletion, and in addition, there are undesirable peaks in the grid [9]. Previous studies, on the one hand, have either contributed to the development of smart charging strategies or isolated BMS algorithms based on A.I., but on the other hand, a continuous gap exists: no single integrated real-time architecture has therefore been established that brings together an AI-driven on-board BMS, a grid-aware smart charger, and a cloud scheduler, all of them working together to jointly optimize battery health, charging cost, and grid stability [10]. The present paper contributes to that gap by suggesting and simulating an end-to-end IoEV framework in which an AI-BMS, bidirectional charging (V2G), and a real-time grid-aware charging station collaborate to the following: (i)

Customize charging according to battery state (SOC/SOH/thermal),

The research finds its basis in the conception, simulation, and assessment of an end-to-end smart charging and battery management framework for IoEVs that promotes system-level energy efficiency and battery life. MATLAB/Simulink has been used primarily as a modeling platform. The smart charging station is integrated with power dynamic load balancing and charging prioritization, both concerning feedback received from the grid in real time; simultaneously, the intelligent BMS module is modeled to conduct advanced SOC/SOH estimation, thermal management, and adaptive charging control.

The objectives of this research are as follows:

Modeling a smart charging station based on priority scheduling and load-aware power regulation.

Studying the EV-smart grid interaction, particularly emphasizing how coordinated charging may help to avoid demand spikes and promote energy sustainability.

There are some major gaps, first gap is to field deployment persist due to varying charger reliability and disproportionate infrastructure costs that work against the adoption of smart chargers. Second, these AI BMSs are often subjected to testing in simulated or isolated environments and are kept separate from real-time grid conditions. Third, smart-charging frameworks exist but are formulated under the assumption of perfect grid communication and renewable forecasts, thereby weakening their unsuitability for real-world operational scenarios. Lastly, though protocols such as OCPP and ISO 15118 appear promising from the standardization point of view, their implementation to date has revealed potential cybersecurity threats that call for a resolute resistant communication framework [19].

Though the literature of 2021–2025 provides important evidence for smart charging, BMS optimization, and grid-interfacing strategies, it is missing an integrated model on hardware that bridges AI-based BMS, smart charging, real-time grid scenarios, and cybersecurity.

The proposed architectural design is meant to integrate the smart charging station with an intelligent battery management system in a real-time data exchange coordination manner within the IoEV framework. The very core of the design is context-aware adaptive charging depending on the status of the battery of the vehicle, grid availability, and dynamic energy pricing [20]. This system comprises electric vehicles furnished with AI-enhanced battery management modules, charging stations with load control capabilities, and a cloud platform for global optimization.

Thus, the architecture allows for the bidirectional exchange of information among EVs, smart charging infrastructure, and a centralized energy management server. In such settings, the EVs transmit real-time battery parameters such as SOC, SOH, temperature, and charging requests. Consequently, the smart charging station considers this information in conjunction with grid status and user preferences to determine the charging profile that optimizes all considerations [21]. Optimization of the said structure further enhances energy efficiency, grid stability, and battery lifetime.

Connecting EVS, smart charging stations, a cloud-based decision engine, and the utility grid are shown in Fig. 2. Dynamic interplay within this IoEV ecosystem is shown, with arrows from exchanged data parameters (SOC, pricing signal) [22].

The AI-BMS is the operational bridge between vehicle-level battery protection and system-level grid objectives. Concretely, onboard SOC/SOH estimators and thermal monitors provide per-vehicle constraints and health metrics that the smart charger and cloud scheduler use to generate charging setpoints that respect both user requirements and feeder limits. Conversely, the cloud and charger provide grid status and economic signals (feeder loading, ToU price, demand-response requests, V2G calls) that the BMS uses to adapt current limits and thermal derating policies dynamically. Through this two-way information flow high-frequency battery protection at the edge and lower-frequency grid/economic coordination in the cloud the BMS design directly implements the architecture’s objectives of (i) preserving battery life, (ii) minimizing charging cost, and (iii) reducing grid peak stress while enabling safe V2G operations.

These smart charging stations are tasked with adaptive control of charging current and voltage to the connected EV. Charging current is computed with regard to the battery voltage of the vehicle and the instantaneous available power from the grid. The equation follows [23]: (1)Icharge=min(Imax,PavailVbat)

In this equation, Icharge denotes as the charging current to be applied, Imax as the maximum allowable current as per the battery specification, Pavail as the real-time available power from the grid, and Vbat as the current battery voltage.

To support multiple vehicles and prioritize charging, the system employs a weighted demand-scheduling algorithm [24]: (2)Psched(t)=∑i=1Nwi⋅di(t)here, Psched(t) represents the scheduled power distribution at time t, di(t) is the power demand of vehicle iii, and wi is the priority weight assigned to each EV, based on charging urgency or user preference.

Real-time communication is maintained via lightweight IoT protocols such as MQTT or HTTP REST APIs, ensuring continuous data flow with minimal latency among EVs, charging stations, cloud servers, and utility grid interfaces. Vehicles communicate the status of charge (SOC), state of health (SOH), internal temperature, and charging status, while cloud servers convey the most optimized charging schedule regarding grid conditions and electricity tariff structures that the smart charger then uses to adjust charging behavior dynamically [25].

Fig. 3 illustrates the flow of data among EVs, smart chargers, the grid, and the cloud. Some of the important parameters-SOC, SOH, grid power availability, and price information-have been inserted along the arrows to emphasize the reactive nature of the architecture.

The AI-based BMS embedded in each EV carries out real-time monitoring and makes it possible the predict battery metrics. SOC is estimated through a modified Extended Kalman Filter (EKF) [26]: (3)SOCk+1=SOCk−η⋅Ik⋅ΔtQnomhere, η denotes as the Coulombic efficiency, Ik is the measured current, Δt is the time step, and Qnom is the nominal capacity of the battery.

SOH is predicted using a machine learning-based mapping function [27]: (4)SOH=f(Voc,Rint,Tcell)here, Voc represents the open circuit voltage, Rint is the internal resistance, and Tcell is the cell temperature. These inputs are processed using regression models trained on historical battery data to predict degradation trends and recommend protective charging strategies.

The fundamental importance of the design and simulation process can be understood by the fact that the principles illustrated in the Table 1 are pivotal to the smooth process [28]:

These assumptions were encoded within the Simulink models to offer realism in simulation outputs with model tractability.

This system architecture gives the structural backbone to a dependable, intelligent, and efficient IoEV ecosystem. With an integrated design for the smart charging station and the AI-based BMS, scalable energy optimization, grid support, and battery life improvements can be realized. Next, we will discuss the detailed design and simulation of the Battery Management System [29].

In the electric vehicle realm and particularly in the Internet of Electric Vehicle setting, the Battery Management System assumes a paramount role in enabling the real-time monitoring, control, and communication over the system. The BMS is concerned with the following major goals for the present study:

SoC Estimation—to properly track the available battery energy for operational reliability.

The BMS interacts with the vehicle’s thermal system and charging controller and accordingly switches between strategies based on the real-time sensor data, historical data, and external commands from the smart charging station [30].

The connectivity shown in Fig. 4 is between the battery pack, sensors, BMS controller, thermal management module, communication links, and the charging station. Sensor blocks may include voltage, current, and temperature sensors that convey information into the estimation algorithms.

The control logic of the BMS operates on the basis of safety thresholds, charging efficiency targets, and real-time inputs. It aims at keeping charging and discharging within the allowable operating conditions [31]: ∙Vmin≤Vcell≤Vmax ∙Tmin≤Tcell≤Tmax ∙Icharge≤Imax

Charging is permitted only when all of the above conditions are satisfied. Thermal derating is applied by the BMS if the cell temperature approaches the upper limit, thereby reducing the charge current: (5)Iderated=Imax⋅(1−Tcell−TthreshTmax−Tthresh),Tcell>Tthres

This thermal regulation ensures the safe operation under variations of ambient conditions.

Two approaches were explored for BMS decision-making: a Rule-Based Logic and an AI-enhanced Controller.

The rule-based logic methodology uses lookup tables and conditional statements derived from design considerations to obtain safe operating conditions and adjust current limits to those values. This is a simple approach; however, newer applications and degradation have not been taken into account.

In contrast, the AI controller uses a regression model trained on battery degradation data. The model predicts SOH from voltage, internal resistance, and temperature [32]: (6)SOH=f(Voc,Rint,Tavg)

This model has been implemented through neural architecture in MATLAB’s Machine Learning Toolbox. It thereby promotes predictive maintenance and condition-based control, primarily in the fast-charging context.

Accurate SOC estimation is done using an Extended Kalman Filter (EKF) formulation. The battery consists of a model representing an equivalent circuit with OCV, series resistance, and parallel RC elements.

The SOC updates the equation in discrete time is evaluated as [33]: (7)SOCk+1=SOCk−η⋅Ik⋅ΔtQnom

The state prediction and correction steps, which follow the EKF framework, are given as [34]:

For Prediction: (8)x^k|k−1=f(x^k−1,uk−1)

To Update: (9)x^k|k=x^k|k−1+Kk(yk−h(x^k|k−1))here, x^ denotes as the SOC estimate, Kk as the Kalman gain, yk as the voltage measurement, and h(⋅) as the nonlinear measurement function.

The EKF algorithm offers the best trade-off in terms of reactivity and noise immunity, which is rendered into actual performance improvements against smart counting only in cases of dynamic loads.

The whole BMS was simulated in MATLAB/Simulink. Real-time interactions are modeled between various subsystems: battery pack modeling, sensors, and control logic.

The battery pack is modeled by a second-order equivalent circuit, considering temperature dependence and degradation parameters. Current and voltage sensors are considered to have noisy data and delay, simulating real conditions. Using the controller subsystem, the estimation of SOC/SOH is carried forward, temperature is checked, and adjustments are made to the charge current.

Communicating data to the smart charging station model creates a feedback-controlled charging environment.

The simulation setup of BMS, shown in Fig. 5, is the BMS subsystem in Simulink, displaying the battery, sensors, estimation blocks (EKF and SOH), and output control logic. The arrows show the real-time data flow.

The key parameters considered in the BMS model are given in Table 2.

The above points dictate the simulation logic and are based on real-life EV specifications.

A BMS thus designed takes care of the dynamic behavior of the battery pack through the sophisticated details of estimation and control algorithms. This allows safe operation and, through its association with the IoEV ecosystem, condition-based intelligent charging.

The smart charging station is engineered to operate dynamically, taking into account prices, user preferences, and grid conditions to ensure the supply of power to connected electric vehicles in a manner considered efficient in the presence of cost constraints and grid availability. The station implements time-of-use pricing, whereby electricity cost varies with time [35]: (10)C(t)={Cpeak,if t∈peak hourCmid,if t∈mid hoursCoff,if t∈off–peak  huurs

The smart charger endeavors to prioritize charging during t∈offpeak hours. This is to minimize cost and grid burden, thereby placing the smart charger firmly with the other elements of demand response.

Differentiated energy allocation is guaranteed by 3 user-allowed charging modes-Fast, Normal, and Eco.

Fast Mode: Maximum permissible current; priority weight w = 1.0.

Power allocation for a vehicle i at time instance t, based on the weighted fair sharing algorithm [36]: (11)Pi(t)=wi⋅Ptotal(t)∑j=1Nwjwhere Ptotal(t) denotes as the total available power and N is the number of connected EVs.

The charging station interfaces directly with the distribution grid and abides by constraints imposed by grid conditions. During peak hours of demand, load balancing considerations require a reduction in charging current [37]: (12)Pgrid(t)+∑i=1NPi(t)≤Plimit

This dynamic constraint stays against feeder overload. On the other hand, the charging station could receive demand response signals from the utility and hard throttle or pause lower-priority charging interruptions.

The model may be implemented to support renewable-energy options such as rooftop solar. When Prenew(t)>0, the charger prioritizes electricity generated locally as follows [38]: (13)PEV(t)=min(Pdemand,Prenew(t)+Pgrid(t))

Furthermore, this charger is capable of bidirectional charging (V2G), allowing, with the stored battery energy, to inject power into the grid [39]: (14)Pnet(t)=PEV→grid(t)−Pgrid→EV(t)

Hence, the bidirectional operation is activated under the utility request and the battery SOC threshold SOC>SOCminV2G. This provides grid flexibility in times of shortage.

The creation of two-way flow diagrams, as in Fig. 6, defines flows with electricity onto charge batteries and to move from them for discharging. Thresholds are defined for switching between modes.

The station is designed to work efficiently by implementing smart charging timing, load balancing, and V2G support. It is a confirmation that the station is responsive to dynamic pricing and multi-vehicle coordination in MATLAB/Simulink modeling is given in Table 3.

We formalize the smart charging controller as a constrained, multi-objective optimization solved in receding horizon (MPC) fashion. Let N be the number of EVs connected at the station and t discrete time indices with step Δt. The decision variable for each EV i is the charging power Pi(t) (positive for charging, negative for discharge/V2G). The model comprises (i) grid and charger constraints, (ii) battery dynamics, (iii) operational limits, and (iv) a scalarized multi-criteria objective.

Grid and charger constraint. At every time step the total station draw must respect the feeder limit [40]: (15)∑i=1NPi(t)≤Pgrid_max(t)∀t

SOC dynamics. The SOC of EV i evolves as: (16)SOCi(t+1)=SOCi(t)+ηiPi(t)ΔtCiwhere Ci is nominal capacity (kWh) and ηi is round-trip efficiency (dimensionless). SOC bounds enforce safety and user targets [41]: (17)SOCmin≤SOCi(t)≤SOCmax,SOCi(Tdeadline)≥SOCi,target

Charging current/power limits and thermal derating. Each EV has a power limit [42]: (18)Pmi,i≤Pi(t)≤Pmax,i(Tcell,i(t))with Pmi,i reduced by a thermal derating function when Tcell,i exceeds threshold Tth [43]: (19)Pmax,i(T)=Pmax,i0⋅(1−kmax(0,T−Tth)/(Tmax−Tth)

SOH degradation proxy. Rather than requiring proprietary SOH curves, we adopt a practical proxy linking short-term degradation rate to average C-rate and temperature [44]: (20)ΔSOHi≈λ1C¯i+λ2T¯iwhere C¯i and T¯i are windowed averages and λ1, λ2 are calibration constants (estimated from lab/field data). This proxy allows the controller to penalize aggressive charging without direct OEM SOH disclosure.

Multi-objective scalarized optimization. We minimize a weighted cost combining peak load, user energy cost, and battery health preservation [45]: (21)minPi(⋅)J=α⋅maxt(∑iPi(t))⏟PeakLoad+β⋅∑t∑iC(t)Pi(t)Δt−γ⋅∑iSOHretiwhere C(t) is time-of-use price, and SOHreti is expected SOH retention (approximated through the proxy). The scalarization weights α, β, γ permit tuning (e.g., utility-centric vs. user-centric). Constraints above (grid, SOC, power limits, safety) are enforced for all t.

The resulting problem can be cast as a convex quadratic program (QP) under linearized SOC and proxy constraints or as a mixed-integer QP when discrete port allocations or mode selection (Fast/Normal/Eco) are modeled. We implement a receding-horizon MPC solved every Δt with horizon H (e.g., 15–60 min) using a commercial QP solver (e.g., Gurobi/OSQP). For large fleets, we propose a distributed ADMM decomposition: each charger/edge node solves a local QP and exchanges aggregate power with a coordinator, preserving real-time performance and reducing central load [46].

Using simulation software MATLAB/Simulink R2023b, the modeling and testing of the control system were carried out, which allows simulation of the dynamic interactions of the electric vehicle, smart charging station, as well as the grid through discrete-time control logic. Modular subsystems for Battery Management System (BMS), smart charging stations, and vehicle interfaces have been constructed through the block-based architecture. For time variation simulation, a multi-rate sampling technique has been used with estimation updates occurring every 50 ms, whereas the charging control logic occurred at 100 ms intervals.

A battery pack whose simulation design represents a mid-size EV having lithium-ion (NMC) chemistry. Internal resistance and temperature behavior are taken into consideration by a second-order equivalent circuit model. The nominal voltage of the battery was taken as 350 V with a 60-Ah capacity. Internal resistance was counted to be 25 milliohms, and temperature evolution was accounted for by a lumped thermal model.

Each of the three drive-cycle load patterns was applied to simulate a different usage condition-urban, suburban, and highway. Each of the profiles consists of phases of acceleration, deceleration, idle time, and regenerative braking. Such scenarios offer a realistic setting where load demand varies and charging frequency, as well as thermal dynamic behavior changes.

Charging sessions included three selectable modes: Fast, Normal, and Eco. These profiles change current limits and are selected based on user priority and grid conditions. Charging logic varies dynamically depending on the current SOC of the vehicle and modes.

This charging station was fed by a grid that had its operational constraints. The maximum power drawn was 200 kW valid maximum. Voltage matrix fluctuations could be tolerated within an amateurish limit range of ±5% about 415 V. Supports for renewables were incorporated on an option basis by simulating photovoltaic availability and allowing grid power with solar input to share in any way possible. For bidirectional flow of power, enhancement of the V2G operation was activated only when the SOC was above the set limit. Detail simulation configuration and battery parameters is given in Table 4.

Major simulation blocks interconnect in the following sequence, as shown in Fig. 7 illustrates the EV load models, battery pack with sensors, BMS controller, smart charger interface, and grid integration. The Flow indicates the direction of signals as well as presents the feedback loops for SOC, thermal feedback, and current control.

A base case consisting of traditional constant-current (CC) charging at fixed power levels without any form of optimization was taken into consideration. This scenario stood as a contrast to the smart charger environment based on dynamic pricing signals, priority-based scheduling, and grid constraints.

Fig. 8 compares the SOC profile of a battery charged under a traditional system and the proposed smart charging logic. The smart strategy shows smoother and adaptive charging curves with efficient cutoff near full SOC.

Under similar load and environmental conditions, smart charging demonstrated a reduction in charging time by up to 12.4%, an improvement in SOC uniformity across the vehicle fleet by 9.7%, and a decrease in charging current fluctuations by 18.2%, thereby enhancing thermal stability.

With the introduction of smart charging and a BMS logic that is adapted accordingly, the SOC and the long-time SOH of the battery could be measurably improved. Conventional systems tend to either overcharge or charge with too high C-rates, thereby accelerating battery degradation.

The BMS with AI-based charging control kept the charging in the optimal SOC window (20%–90%) and was also able to control battery temperature by dynamically throttling current.

The degradation curve shown in Fig. 9 with 100 charge-discharge cycles indicates a slower capacity fade with the smart charging system, accounting for nearly 92% SOH against the 85% SOH of the conventional one. Eight-point-one percent was recorded in SOH drop after 100 simulated cycles for the smart charging system, while the traditional one recorded 14.6%. This conversion results in an evaluated extension of approximately 20% in the life of the battery by way of the smart charging method.

The charging station has the strength of optimizing charging during off-peak hours and cutting down on overall energy use. The algorithm acts on the potential inherent in time-of-use prices to reap tangible benefits for the energy economy and grid stability.

The energy cost per vehicle would be down with off-peak optimization: This is the 24-h cycle of total grid load in Fig. 10, showing the sharp peak demand under traditional charging and the flattened profile achieved with smart scheduling.

Peak load reduction by up to 38% under smart charging meant the demand curve was effectively flattened and eventually contributed to strengthening grid reliability, as shown in Table 5.

Allowing smart scheduling of energy flow, power was allocated dynamically based on user mode selection (Fast, Normal, Eco) and grid availability. Fast Mode vehicles were given preferential treatment, and Eco Mode ones were deferred when the grid experienced stress.

The selectivity efficiency is defined as the energy delivered to the battery relative to the total input energy, inclusive of losses. (22)η=EbatteryEinput×100

Results revealed a smart charging efficiency of 92.6% vs. 89.3% for the traditional system. Time-wise, charging times were optimized as well; in Fast Mode, the smart system finished charging in 1.3 h, while the traditional method took 1.5 h to charge. Eco Mode occurred when the smart system took 3.8 h to charge, with energy loss reduced by 26%, showing that Arduino-based EV algorithms perform better in terms of energy management, as shown in Table 6.

The system was tested in scalable conditions with up to 10 EVs attached simultaneously. A communication delay of up to 200 ms was imposed to assess its response time under real-world conditions.

The smart charging controller proved to have real-time capability, whatever that means; decision latency was less than 80 ms, response time to grid signals was less than 150 ms, and maximum simulated CPU utilization was 63%.

Hence, this confirms feasibility for real-world deployment using microcontroller or embedded edge devices with moderate computational resource requirements.

Batteries usually degrade much faster than a regular consumer use due to charging at high temperatures. However, the very presence of a charge current far above normal can induce temperatures harmful to this very electrochemical conversion process. The BMS thermal module dynamically reduced charging current when its internal temperature exceeded 43.1°C. This can give a temperature drop of the battery peaks of up to 47.5°C in comparison to a conventional charging approach. So, this greatly improved the thermal reliability and safety rating of the battery.

Fig. 11 shows that Smart charging kept the maximum temperature below 43.1°C, whereas traditional methods led to peaks over 47°C, also listed in Table 7.

The need for thermography is recognized as an important condition for the BMS system.

The smart charging station balances the charging of unfolding EVs, each having a different priority. The controller for load balancing schedules depending on the mode (Fast, Normal, Eco), on SOC, and on whether grid power is available.

As comparing modes between the fast and economy, the pulls from the batteries are imperfectly loaded, then more reserve toward Eco-mode batteries during peak demand times, as shown in Table 8. Power draw of individual EVs under smart scheduling is shown in Fig. 12.

We confirm that the Intelligent Controller perfectly balances user preferences and grid constraints in real-time.

V2G activation was also possible in grid emergencies. Whenever large capacity V2G e-generators were implemented for grid services, high capacity V2G e-compressors would also be permitted to be implemented anytime and anywhere.

The EVs started collectively discharging up to 40 kW back into the grid during 6–8 PM to support voltage and frequency stability (See Fig. 13).

Key observations drawn from the simulation include a V2G response time of less than 200 ms, voltage regulation within ±2%, the power factor greater than 0.96, and roughly 10.3 kWh of energy returned per EV given in Table 9. The simulation once more confirms that the bidirectional charger supports grid resilience and effective peak shaving.

In comparison to traditional systems, the smart charging station integrated with the AI-based BMS is advantageous. The key benefits are speedy and efficient charging, minimum battery wear and tear, cheaper energy cost, and better compliance with grid conditions. These augmentations make a step towards making the Internet of Electric Vehicles (IoEVs) more sustainable and scalable.

Being real-time feasible, the proposed model encourages further experimental validation and implementation through smart cities. The integration of renewables with predictive analytics will lend itself to further enhancements in future versions.

Cybersecurity is a critical enabler for any IoT-enabled charging infrastructure because attacks can directly affect battery health, user safety, and grid stability. Threats most relevant to the proposed architecture include: (i) false telemetry (spoofed SOC/SOH or temperature reports) that could trigger unsafe charging profiles or incorrect V2G discharge; (ii) man-in-the-middle (MitM) attacks that alter price or demand-response signals to profit an attacker or to intentionally stress a feeder; (iii) denial-of-service (DoS) attacks on edge/charging nodes that undermine coordinated scheduling and result in uncoordinated charging surges; and (iv) firmware compromise of on-board BMS/charger controllers that can modify charging limits or disable safety interlocks.

Practical mitigation for this architecture should combine lightweight cryptographic authentication (mutual TLS/ISO 15118 Plug-and-Charge style credentials), secure boot and signed firmware for edge devices, end-to-end message integrity checks (HMAC or similar), redundancy of critical telemetry (cross-validation with charger-side voltage/current measurements), and anomaly detection at both edge and cloud layers (statistical or ML-based monitors that flag improbable SOC/SOH reports or abrupt changes in charging patterns). For transactions which require ledgered auditability (e.g., energy transfers in the V2G scenario), selective blockchain-style logging techniques or evergreen ledgers can be employed for forensic analysis and non-repudiation while keeping the operational path lightweight.

The inclusion of these security features in the edge controllers that are resource-constrained demands balancing the security overhead, latency, and energy consumption. Therefore, we recommend a layered approach: critical control messages and authentication use low-latency, hardware-accelerated crypto; less time-sensitive logging can be handled asynchronously; and anomaly detectors operate locally to avoid single points of failure. Future prototype work (HIL and field pilots) should include security testing (red-team exercises and penetration testing) to quantify latency and operational impacts of the chosen security stack.

Various challenges arise in the practical deployment of IoT-based charging stations which grossly stem from the handling of proprietary data of vehicle-side BMS. In practice, charging stations have limited access to detailed states of charge or state of health, or priority indicators, as these parameters are usually locked away within the OEM’s proprietary framework. With this dependence, deploying the schemes in the real world becomes quite challenging when the proposed framework explicitly assumes access to granular internal states.

The proposed framework tackles this limitation, so it aims to minimize a lot of dependence on unstandardized BMS data and would, instead, focus on standard and, from outside, measurable signals. Currently, communication scenarios such as ISO 15118 (Plug-and-Charge) and OCPP 2.0.1 support standardized SOC reporting, interoperability, and secure data exchange. In the absence of direct reports of SOC/SOH, estimations can be made by utilizing charger-side measurements of current, voltage, and temperature in the context of an Extended Kalman Filter, an electrochemical impedance model, or an AI-based predictive model. Moreover, demand flexibility can be obtained at a user level by implementing a priority-linking charging interface expressed as a user-declared priority system (Fast, Normal, Eco), without requiring the OEM to keep proprietary priority flags internally. On a real-world basis, hence actual functionality and scalability of such systems is required, alongside considering the directives-based protocols, external estimation models, and user-defined restrictions to support intelligent and optimized charging and V2G switching.

While the system proposed has shown great promise from the simulation perspective, it needs to be further revised for practical deployment to come to the fore. One of the first things to be done is the establishment of a Hardware-in-the-Loop (HIL) simulation environment. This will allow for the real-time validation of the smart charging controller and the BMS algorithms by coupling the simulation of EV batteries and grids to the actual control hardware. This will measure latency, control precision, and robustness of such a system in real-world dynamic situations.

A promising extension is evaluation of the proposed IoEV framework within wholesale/retail electricity market contexts, where EV fleets participate in market transactions (day-ahead/real-time markets, ancillary service bids) and are scheduled with economic dispatch constraints. Future work should evaluate market-participation strategies (e.g., revenue stacking from energy, capacity and ancillary services) and the trade-offs between revenue, battery degradation and reserve requirements. Recent work demonstrates the potential for economic dispatch of EVs in electricity markets and provides modeling approaches and case studies for quantifying these trade-offs (see e.g., Appl. Energy, 2024, DOI:10.1016/j.apenergy.2024.124347). Incorporating such market models will allow assessment of the full techno-economic value of AI-BMS + V2G coordination under real market price dynamics.

The construction of a working prototype of the smart charging station is also in the pipeline. This prototype will incorporate embedded controllers, smart metering hardware, and user interface modules. Testing of scheduling logic and flow of communication between the EVs, the charging unit, and the cloud server, as well as interoperability with different EV models, will be done on the prototype [52].

To add further scalability and dynamic response, the cloud charging scheduler will be developed. It shall collect the grid status in real time, such as load demand, frequency, and tariff data, in order to adjust the charging strategy. Coordination at the fleet level of EVs shall be facilitated, which is important.

Lastly, it is crucial to address security challenges in an IoT-based EV infrastructure. Future research would thus develop a secure communication layer: encryption protocols, anomaly detection algorithms, and blockchain-based logging mechanisms. Data integrity must be ensured at all times, and device authentication must be performed securely to preserve the privacy of the end-user and instill trust in public charging networks.

Altogether, these future developments will carry the system from a mere simulation model toward a fully deployable solution to next-generation smart mobility.