SLICED: A secure and adaptive cloud–iot framework for low-latency e-learning environments

Abstract

Providing dependable, secure connectivity remains a persistent challenge in digital education, particularly in data-sensitive, remote learning environments. This study presents SLICED, which stands for Secure Learning Integration via Cloud and Edge Devices. It is a framework that integrates Internet of Things edge devices with Amazon Web Services (AWS) Cloud services. SLICED orchestrates AWS IoT Core, Lambda, and Key Management Service (KMS) to enable encrypted communication, user authentication, and real-time edge analytics. When compared to traditional AWS–IoT educational systems, this adaptive integration cuts down on latency and increases the level of data protection. The results of experiments conducted in simulated learning networks demonstrate that SLICED can achieve up to 27% lower latency and 33% greater security, thereby providing smart learning environments that are both scalable and safe.

Introduction

In view of the rapidly changing state of education, the convergence of cloud computing and the IoT represents a paradigm shift toward safe, interconnected learning environments¹. Traditional learning systems have struggled with scalability issues such as data breaches, limited flexibility, excessive latency, and centralized designs². There are many challenges affecting environments with limited bandwidth or long distances within the geographical area³. During the end-user digital learning experience, these barriers hinder their ability to perform digital learning once important data is engaged or access is required in real-time⁴. With the potential of AWS Cloud and IoT-enabled edge computing, this research offers a secure and extensible learning framework that addresses these limitations⁵. While not losing sight of the fundamental purpose of digital education, the design of SLICED provides robust data protection, reliable connectivity, and adaptive resource allocation⁶. Unlike conventional systems that centralize data processing, SLICED combines AWS services with local edge processing to enable intelligent, secure data processing, encryption, and transfer⁷. This approach increases responsiveness, reduces latency time, and enhances data security⁸. The proposed framework addresses the performance and security deficiencies of conventional modes while retaining the instructional intent through cloud-based intelligence and real-time adaptability⁹.

Problem statement

Particularly in distant and remote settings, current learning systems struggle to deliver low-latency, scalable, and secure performance¹⁰. Effective and timely learning experiences are hampered by their poor scalability, restricted real-time responsiveness, and inadequate data protection. Furthermore, most current technologies lack an integrated architecture that ensures reliable data protection and seamless connectivity between the cloud and the Internet of Things. SLICED integrates AWS Cloud services with Internet of Things and edge computing to provide safe, real-time, and scalable learning. SLICED improves data security and resilience with AWS IoT Core and KMS, while edge computing integration reduces latency and facilitates real-time interaction. Additionally, its cloud-based automation and intelligent data management maximize resource usage without compromising training. SLICED quantitatively improves latency, scalability, and response efficiency, making learning more dependable and responsive.

Contributions of this paper

The major objectives of this paper are;

• SLICED platform enhanced the power of AWS Cloud along with the Internet of Things to allow for real-time response. This capability enables learning systems to respond in real time to changes in the environment and user interactions, ensuring a reliable and useful training experience.

• Using edge computing in conjunction with secure cloud services, like AWS KMS and IoT Core, helps to shore up platform security. In addition to facilitating faster, more secure learning processes, edge computing reduces latency by reducing the need to be constantly connected to the cloud.

• SLICED utilizes cloud-based automation and smart data management to ensure the proper use of system resources. The increase in efficiency is achieved without diminishing the quality of instruction or the dependability of the platform. The focus remains predominantly on the educational goals.

AWS IoT and edge computing modules were integrated to reduce latency and provide adaptive learning interactions for real-time responsiveness. AWS KMS and secure edge-cloud communication layers were chosen based on data security. To balance cloud and edge workloads, cloud-based automation and smart data management were used to maximize resource consumption. These objectives provide the design rationale that enables SLICED to meet its scalability, security, and performance requirements.

The remaining section of this paper is organized as follows: Sect. 2 reviews past studies on ensuring secure and seamless connectivity for learning systems, which is critical. Section 3 describes the proposed SLICED process. Section 4 compares our suggested approach with other conventional methods. Section 5 concludes with a discussion of potential future studies.

Related works

This literature review examines security in the educational environment, cloud computing, Internet of Things integration, and edge computing. This report identifies the limitations around data security, latency, and scalability. The findings help strengthen the proposed SLICED framework by showing how to leverage AWS, Cloud, and Internet of Things integration to fill in the missing pieces and enable real-time connected learning spaces, while preserving students’ personal information.

Cloud computing in education

This article discusses cloud computing in educational institutions and offers recommended steps for a multi-layered cloud adoption approach to improve content delivery and scalability¹¹. While the study outlines other issues, such as privacy concerns and infrastructure readiness in underdeveloped contexts, the article identifies significant benefits of cloud implementation, including increased data access and collaboration. This paper examines how the IoT, cloud computing, and online learning, among other technologies, have been amalgamated in modern-day educational settings. With an emphasis on increasing accessibility and usability, it proposes a unified architecture that incorporates these technologies¹². Study indicates that this connection enhances the effectiveness of learning; however, there is minimal, general acceptance due in part to questions of security, and reliable connectivity and infrastructure.

IoT integration in smart learning systems

The research provides a conceptual structure for interactive learning experiences that support the use of sensors and Internet of Things gateways with an emphasis on smart learning through the Internet of Things (SL-IoT). The proposed approach enhances student engagement and improves real-time monitoring¹³. While results indicate that learner tracking was improved and responsiveness was increased, issues with data privacy and device interoperability remain. This article discusses LearnSmart, a platform that connects LMSs to the IoT (LMS-IoT). It changed the lessons based on the sensor data and the comments it received about the situation¹⁴. The results revealed that using the system, doing well in school, and being motivated were all linked in a good way. Data security and scalability needed to be looked at.

Security and privacy in cloud-based learning

The study examines privacy and security issues in cloud-based or course-based online or e-learning systems. The study proposes a policy enforcement model based on encryption (PEF-En) through data classification and identity management¹⁵. The outcomes showed less exposed data, and better compliance are possible; however, the study points out that there are significant real-world issues related to performance trade-offs, and challenging implementation. Although this study focuses specifically on e-health, it still offers useful insights for educational cloud security¹⁶. It proposes a role-based encryption access control (R-BEC) that meets users’ privacy requirements. The findings indicate improved confidentiality and restricted data access with no additional computational overhead, which is terrific for an educational cloud system.

Edge computing for low-latency educational applications

This paper investigates how to mitigate latency in cloud workloads by converging artificial intelligence (AI) with edge computing¹⁷. The paper proposed an integrated hybrid model of edge deployment methods and AI-based task prediction (H-Ed-AI). The results are faster processing and responsiveness, opening the door to innovative education systems and other real-time applications. For very reliable, low-latency connectivity, the authors propose a solution called mobile edge computing and their approach relies on adaptive task offloading and resource slicing¹⁸. This is especially important for real-time educational systems, which need stability in connectivity and performance, since the results show the drastic improvement in latency and reliability.

Challenges in traditional E-learning arch

This review provides an overview of cloud-based e-learning systems (Cc-ELn) and the most commonly observed features, barriers, and future potential. It proposes a safe and scalable infrastructure from a cloud-based e-learning perspective¹⁹. The research concludes that cloud-based e-learning systems are flexible and more cost-effective, despite ongoing challenges, including integration and cybersecurity issues. The study investigated problems with online education from a myriad of perspectives, ultimately separating them into three categories: technical, cognitive, and contextual. It proposed a mapping model for learning systems²⁰. Despite adaptive systems improving engagement and outcomes, the research suggested that integrating customization into cloud platforms remains difficult.

The studies analyzed what the articles found to be the positives and negatives of cloud, IoT, and edge technology in the classroom. There have been examples where improved response times and accessibility have benefitted outcomes; however, challenges related to privacy, scalability and latency continue. The SLICED framework was developed with cloud-edge synergy intended to produce learning experiences that are safe, individual, and timely, with the best of education’s basic tenets in mind, as this set of studies demonstrates. Table 1 shows the summary of related work.

Table 1 Reviews of recent cloud, edge, and hybrid innovative learning and linked environment frameworks. The table organizes references by operating environment, technique type, security, and energy efficiency. This organized analysis presents safe, efficient research trends, strengths, and trade-offs in educational and IoT systems.

The design of safe and flexible data frameworks for various intelligent environments has changed as a result of recent developments in edge and cloud computing. Using edge computing, Yao et al.³⁰ presented a framework for biometric privacy protection in UAV-based systems, enabling safe data processing at network boundaries. Similarly, focusing on decentralized security methods, Yao et al.³¹ presented a privacy-preserving data collection paradigm for intelligent edge systems. Dong et al.³² developed a blockchain-assisted self-sovereign identity solution to ensure transparent user identity management and reliable authentication in UAV delivery networks. Additionally, Yao et al.^33,34 addressed integrity and privacy in distributed settings by designing a comprehensive security architecture for edge computing infrastructures.

Federated learning approaches such as FedShufde, which protect sensitive data while facilitating collaborative edge learning, were investigated to improve privacy and scalability. This was expanded by Dong et al.³⁵, who used task distribution and blockchain technology to secure UAV communications, enhancing energy efficiency and privacy. MoCFL, a mobile cluster federated learning architecture for extremely dynamic edge networks, was presented by Fang et al. to increase adaptability and latency management³⁶. The suggested SLICED framework for educational settings is conceptually grounded in these studies, which collectively provide a solid foundation for flexible, private, and secure edge–cloud integration.

Methods

The proposed methodology contains SLICED system which can make a smart learning environment that is safe, scalable, and works in real time by combining IoT devices with AWS Cloud services. It fixes problems with traditional systems by making data more secure, reducing latency, and enabling flexible content distribution. The design ensures end-to-end connectivity without losing sight of important teaching goals.

Research Hypothesis.

• The IoT–edge–cloud smart learning architecture will increase student engagement and learning results statistically compared to a traditional LMS without IoT integration. ​.

• Edge-layer preprocessing and filtering of multimodal sensor data will minimize network bandwidth and end-to-end latency, allowing quicker user interface learning content adaption.

• ​ Using managed cloud services like AWS IoT Core, Lambda, S3/DynamoDB, and KMS will deliver scalable learning analytics with high data security, integrity, and availability. ​.

• Continuous input from learning analytics to instructors and learners will improve customization and data-driven decision-making, enhancing course completion rates and learner performance metrics compared to baseline offers.

SLICED system overview

The layered architecture, combining IoT with edge processing and the AWS cloud administration, enables efficient, scalable, and secure innovative learning in our proposed system SLICED. Preprocessing the data at the edge layer reduces latency and filters out excess data. IoT devices that capture user activity or environmental context record the data in real time. User devices deliver data to an edge layer for filtering and safe transmission. AWS IoT Core manages connectivity, AWS Lambda processes AI-driven events in real time, and AWS KMS protects data. DynamoDB and S3 protect user data and learning materials. Adaptive content, administrative controls, and dashboard feedback ensure low latency, scalability, and security in the e-learning environment.

Fig. 1

System architecture of SLICED for secure learning connectivity. The picture depicts how user devices and edge layers safely send data to AWS IoT Core. AWS Lambda and AWS KMS implement safe processing and storage in Amazon DynamoDB and S3. The integrated system’s dashboard shows end-to-end secure communication, data management, and real-time analytics with AWS cloud-edge connection.

In the SLICED architecture, AWS IoT Core, Lambda, and KMS interact in real time (Fig. 1). User devices and edge nodes securely send data to AWS IoT Core, which manages device connectivity and communication. AWS IoT Core automates learning events in real time with AWS Lambda. AWS KMS manages encryption keys to keep data transmission and storage—whether for DynamoDB user records or S3 teaching materials—secure. The platform’s integrated dashboard allows adaptive content delivery, feedback, and centralized management, ensuring low latency, data protection, and responsive learning.

Eq. 1 models how IoT device data is captured $\:\int\:{D}_{cap}$and structured.

$$\:{D}_{cap}=\int\:(rtd\left(sn-ms\right)+bh\left({pr}^{{\prime\:}}-ac\right)\text{*}fd-{ud}^{{\prime\:}}\:\:)\:$$

(1)

IoT devices collect real-time data from sensors $\:rtd\left(sn-ms\right)$ that measure anything from environment to student behavior to physiological responses. Eq. 1 guarantees that data is collected accurately $\:bh\left({pr}^{{\prime\:}}-ac\right)$, and this is critical because correct data is the bedrock of adaptive learning $\:fd$. Preprocessing provided the next step in the arrangement of the data to eliminate the unimportant data $\:{ud}^{{\prime\:}}$and allows the system as a whole to be more efficient through eliminating unnecessary inputs $\:bh\left({pr}^{{\prime\:}}-ac\right)$for the preceding orders.

At the edge layer, Eq. 2 describes how raw data from IoT sensors is first preprocessed $\:D\left(p\right)$prior to being sent.

$$\:{D\left(p\right)=En}_{rs}-{ep}_{rd}\left({cl}^{{\prime\:}}-e{p}^{{\prime\:}{\prime\:}}\right)+bd\left({rl}^{{\prime\:}}-tle\right)+{op}^{{\prime\:}{\prime\:}}$$

(2)

This includes filtering, feature extraction, and compression, all of which are done to eliminate noise and reduce $\:En$ the size of the data $\:rs$ in Eq. 2. By the end of the edge processing stage, the relevant information $\:{ep}_{rd}$is all the data that makes it to the cloud layer$\:{cl}^{{\prime\:}}$. This edge processing $\:e{p}^{{\prime\:}{\prime\:}}$step reduces the amount of bandwidth $\:bd$ used and reduces latency $\:{rl}^{{\prime\:}}$in time-sensitive learning $\:tle$ environments, and optimizes system efficiency$\:{op}^{{\prime\:}{\prime\:}}$.

AWS-managed cloud and edge services are used in the SLICED architecture to process data and deliver adaptive content using machine learning and AI models. Specifically:

• AWS IoT Greengrass and Lambda@Edge deploy lightweight incremental learner state predictors such online gradient-based customization models and federated learning schemes on IoT-enabled devices. These models adjust to user interaction and environment for real-time customisation and feedback without cloud dependence.

• Cloud layer: Recommendation systems, NLP modules, and LLMs—including Amazon SageMaker and Amazon Bedrock—improve content creation, adaptive distribution, and Q&A generation. Federated learning or asynchronous aggregation synchronizes these models with on-device models for scaled intelligence and system-wide optimization.

• Automation and control: Event-driven AI pipelines are orchestrated by AWS Lambda, which processes streaming data, triggers adaptive content generation, and manages intelligent automation workflows for educational tasks. These include personalized recommendation systems, real-time transcription and translation, and interactive assessment tools.

The core model learns baseline behavior using an encoder-decoder architecture trained on normal traffic data. Significant deviations in reconstruction error suggest aberrant activity. Adam-optimized, mean squared error-trained autoencoders use many dense layers with ReLU activation. We use dynamic thresholding based on reconstruction error statistics to adapt to changing network conditions for robust detection. Real-time network metrics data gathering, a visualization dashboard, and model fine-tuning for ongoing improvement are also included. On a large, publicly available dataset, the model outperformed PCA and isolation forests in differentiating normal and abnormal behaviours. This comprehensive solution allows scalable, real-time anomaly detection for trustworthy network management.

Step 1: data acquisition through IoT devices

The SLICED system begins by collecting data from the learning environment using IoT devices. In particular, smart boards that influence engagement instantaneously, sensors to detect elements of the physical environment (e.g., light, motion, temperature), cameras to automatically take attendance or track behaviours, wearable to record physiological response (e.g., heart rate, activity) etc²². Different classroom technologies including smart boards, environmental sensors, and cameras collect real-time educational and environmental data. Sources capture student inputs, environmental conditions, and attendance automation. A microcontroller or gateway like ESP32 or Raspberry Pi timestamps and structures all signals. Before sending data to the processing layer for advanced analysis, anomaly detection, and adaptive learning, the microcontroller layer provides accurate aggregation and preparation. This integration ensures situational awareness, automation, and precise data collecting in smart and remote learning settings.

All of the devices are linked by microcontrollers or gateways (e.g., Raspberry Pi, ESP32) for initial handling and structuring of the device-level data collection as elucidated in Fig. 2. This layer enables learning and digital infrastructure to work together smoothly. It additionally turns on the system’s real-time features. After data is gathered from the sources, it is organized, timestamped, and placed in a queue for later processing. This layer is very important for allowing adaptive learning with as little help from people as possible. It ensures that the preprocessing and decision-making layers receive consistent, well-structured input, and provides a framework to monitor learning behaviour at every level.

Fig. 2

Data acquisition in a smart learning environment. Smart boards, environmental sensors, and cameras collect real-time classroom data, which is structured and timestamped via microcontroller gateways for further processing.

Eq. 3 represents the procedure for obtaining real-time data $\:Rtd$ from IoT devices within the learning environment $\:le{\prime\:}$.

$$\:{Rtd=le}^{{\prime\:}}\left({G}_{d}\left({m}^{{\prime\:}}-sr\right)\right)+{ts}^{{\prime\:}}\left({dc}_{e}Q-A{c}^{{\prime\:}}\right)\text{*}T{a}^{{\prime\:}{\prime\:}}$$

(3)

It establishes how data are gathered considering multiple sensors (e.g., smart boards, motion detectors, wearables) $\:\left({G}_{d}\left({m}^{{\prime\:}}-sr\right)\right)$. The Eq. 3 timestamps$\:{ts}^{{\prime\:}}$ and categorizes the data such that it can be later processed$\:\:{dc}_{e}Q$. It is a well-structured equation to ensure that accurate and timely $\:A{c}^{{\prime\:}}$, labeled data have been collected because accurate and timely data $\:T{a}^{{\prime\:}{\prime\:}}$are the foundation of all further processing, decision-making, and feedback that the system will provide$\:{ts}^{{\prime\:}}\left({dc}_{e}Q-A{c}^{{\prime\:}}\right)*T{a}^{{\prime\:}{\prime\:}}$²³.

Step 2: Edge-level preprocessing and filtering

In Phase 2, data from IoT devices is sent to edge computers. Edge “nodes” (small computers or “smart gateways”) process data in real time at or close to the original data source. The primary function of edge nodes is to discard superfluous data, perform preprocessing (e.g., compression and feature extraction), and perform other functions (e.g., activity classification or sensor threshold determination), and to provide a cache to store data when network conditions temporarily cause slow processing.

Algorithm 1

Data Preprocessing and Edge Filtering.

The algorithm 1 filters managed data at the edge based on similarity scores. The algorithm reviews each managed data instance and compares the computed similarity score to a particular threshold. If the managed data instance has a similarity score above the threshold, it is relevant to the filtered data set and added to the filtered data set instance. If multiple managed data instances have the same best score, all instances will be added to the final results set. This will assist the edge device when sending only the most relevant scored data to the cloud for processing²⁴.

Fig. 3

Edge data processing pipeline for IoT-enabled learning environments. Edge nodes perform filtering, feature extraction, caching, and encryption before securely transmitting compressed data to the AWS cloud layer.

Edge nodes or smart gateways like Raspberry Pi and small PCs collect IoT device data. These nodes rapidly filter incoming data, extract key aspects for analysis, and cache it for spikes or connectivity outages. All acquired data is compressed and organized for storage and transmission. Local data encryption protects sensitive data before leaving the edge. After compression, encrypted data is safely sent to the AWS cloud layer, assuring privacy, efficiency, and integrity along the edge-to-cloud pipeline. This layer enhances system responsiveness, reduces latency, and allows applications to remain operational even in the event of a network outage by reducing the volume of data transferred to the cloud in Fig. 3. Data encryption at the local level is the first step to introducing security mechanisms. The data is organized and filtered at the edge layer before it is securely transmitted to AWS cloud services. Identifying and processing data to ensure that what proceeds to the cloud for analysis and decision-making is clean, relevant, and valuable is more efficient and effective with this mechanism.

Eq. 4 explains the process of filtering irrelevant data$\:{fi}^{{\prime\:}}$ and then compressing this data at the edge layer $\:Dc$, which guarantees the edge will only send relevant, appropriate, and high-value data to the cloud.

$$\:Dc={fi}^{{\prime\:}}\left({T}^{{\prime\:}}-{Og}^{{\prime\:}}\right).{aM}^{{\prime\:}}\left(r{b}^{{\prime\:}}-dq\right)\text{*}ep\left({td}^{{\prime\:}}-me\right)$$

(4)

The overarching goal $\:{Og}^{{\prime\:}}$of this equation relates to reducing the amount of data $\:{aM}^{{\prime\:}}$sent to the cloud, reducing bandwidth used $\:r{b}^{{\prime\:}}$, and reducing the delay $\:dq$ or latency from processing edge data, whether that be analysis of data retrieval with Eq. 4. Feature extraction is another way in which edge processing $\:ep$ can be described as organizing the more relevant components to reduce the total data $\:{td}^{{\prime\:}}$to the most useful features to maximise efficiency $\:me$ in the architecture and real-time decision from the edge²⁵.

The Eq. 5 models the extraction of relevant features $\:Fe$ from raw IoT sensor data at the edge.

$$\:Fe=\left(vp\right)+r{d}^{{\prime\:}{\prime\:}}-avc\left(Sy-d{t}^{{\prime\:}{\prime\:}}\right)\text{*}\:mx\left(dc-bc{l}^{{\prime\:}{\prime\:}}\right)$$

(5)

The previous metrics illustrate how different variables or patterns $\:\left(vp\right)$are identified and ranked. The role of Eq. 5 is to ensure that, the most relevant data $\:r{d}^{{\prime\:}{\prime\:}}$ (e.g., movement patterns, engagement data) is processed or made available for the cloud $\:avc$. In this way, the system $\:\left(Sy-d{t}^{{\prime\:}{\prime\:}}\right)$can suppress unnecessary data clutter and maximize $\:mx\left(dc-bc{l}^{{\prime\:}{\prime\:}}\right)$ the organization’s ability to focus on insights that matter in promoting the process of adaptive learning.

Eq. 6 calculates the reduction in latency $\:lr$ achieved by leveraging the edge layer to evaluate data locally.

$$\:lr={pr}^{{\prime\:}}-m{f}^{{\prime\:}{\prime\:}}.si"(sr-dl{\prime\:}{\prime\:})\text{*}\:dl(fd-ee{\prime\:}{\prime\:})\:\:$$

(6)

The processing and filtering $\:{pr}^{{\prime\:}}$of data close to the creating source minimizes $\:m{f}^{{\prime\:}{\prime\:}}$the time required to send information $\:si"$to the cloud. The aim of Eq. 6 is to improve the overall system’s responsiveness $\:sr$, allowing adaptive learning $\:dl{\prime\:}{\prime\:}$content or feedback $\:fd$ to be produced without delay $\:dl$, a key consideration to keep a learning environment engaging $\:ee{\prime\:}{\prime\:}$.

Step 3: secure cloud processing with AWS services

Step three involves securely processing data in the cloud using AWS services. AWS IoT Core uses TLS/SSL with each connected device to identify devices and ensure the secure transfer of data. AWS provides real-time data processing with AWS Lambda, a serverless compute service that can trigger automatic responses like alerts about performance, content recommendations, or anomaly detection. All stored and processed data utilizes AWS KMS for security and access control.

Edge devices send encrypted data to AWS IoT Core, which authenticates devices using TLS/SSL. Data is processed by AWS Lambda to generate real-time alerts or recommendations. AWS S3 (object storage) and DynamoDB (NoSQL DB) store and event logs, protected by AWS KMS for data privacy. Customized content and secure computation are supported by the integrated stack. This optimises learning system outputs, ensuring reliability, privacy, and adaptive service throughout the educational process. AWS S3 stores educational resources, logs, and multimedia files, whereas AWS DynamoDB stores mutable data, such as access patterns, system configurations, and user activity logs, illustrated in Fig. 4. The SLICED framework’s intelligence resides in the cloud layer, enabling data optimization for personalization, enforced access control, and learning orchestration²⁶. Multiple services are integrated, enabling fast, secure, low-latency decision-making at scale, reliability, and data integrity. All this without compromising the learning environment or learner experience.

Fig. 4

Safe processing and content optimization for smart learning on AWS. AWS authenticates, analyzes, and stores encrypted edge data, with KMS assuring confidentiality and delivering tailored content.

Eq.7 represents the encryption process $\:En$ while transferring data from the edge to the AWS cloud.

$$\:En=\left[ep\left({as}^{{\prime\:}}-{ua}^{{\prime\:}{\prime\:}}\right)\right]+P{r}^{{\prime\:}}-\left\{ssp<{ua}^{{\prime\:}}-us>\right\}\:\:$$

(7)

This is critical to measure the security strength of any encrypted data $\:En$ secured using encryption protocols $\:ep$ like TLS/SSL. It gives the analyst assurance $\:{as}^{{\prime\:}}$sensitive educational and user data $\:{ua}^{{\prime\:}{\prime\:}}$are secure and protected $\:P{r}^{{\prime\:}}$ during the transfer process by Eq. 7. Strong encryption protocols $\:ssp$ will disallow illegal, unauthorized access $\:{ua}^{{\prime\:}}$and data will remain secure and unbreached $\:us$. This preserves the integrity and confidentiality of the system.

Eq.8 illustrates the way in which data is processed $\:dp$ in real time in AWS Lambda.

$$\:dp=sF\left\{ad{q}^{{\prime\:}{\prime\:}}\right\}+c{p}^{{\prime\:}{\prime\:}}<ad-d{q}^{{\prime\:}{\prime\:}}>\text{*}\left\{ps\left(al-ue\right)\right\}$$

(8)

The above Eq. 8 illustrates how serverless functions $\:sF$ enable automatic downstream action $\:\left\{ad{q}^{{\prime\:}{\prime\:}}\right\}$on incoming data like alerts, content updates $\:c{p}^{{\prime\:}{\prime\:}}$or anomaly detection $\:ad$. The purpose of the formula is to process data quickly $\:d{q}^{{\prime\:}{\prime\:}}$and at scale in the cloud, as well as not have to provision servers $\:ps$. It allows users to ensure that adaptive learning $\:al$ features run in real time, thus guaranteeing that the user experience $\:ue$ is seamless and suitable for the user.

Eq. 9 is a model of the cloud’s access control method, based on AWS KMS, the cloud access control $\:ac$ protocol itself.

$$\:ac=\left[pr\left({ud}^{{\prime\:}}-{md}^{{\prime\:}{\prime\:}}\right)\right].s{e}^{{\prime\:}}-pd\left(sc-c{o}^{{\prime\:}{\prime\:}}\right)\:\:$$

(9)

It uses processes $\:pr$ of user and device $\:{ud}^{{\prime\:}}$authentication or access control, to identify what users and devices are permitted to access data and make modifications $\:{md}^{{\prime\:}{\prime\:}}$by Eq. 9. The goal of this equation is to secure educational content $\:s{e}^{{\prime\:}}$, personal data $\:pd$, and system configurations $\:sc$, while simultaneously establishing real control $\:c{o}^{{\prime\:}{\prime\:}}$over who can access what information in real-time $\:pd\left(sc-c{o}^{{\prime\:}{\prime\:}}\right)$, a foundational security and protection feature of sensitive learning data.

Algorithm 2

Dynamic resource allocation for cloud processing.

Algorithm 2 accrues resources in the cloud, subject to the allocation score. The algorithm checks each cloud resource and its allocation score against the prior maximum. If a resource’s allocation score is higher than or equal to the current best score, the resource is included in the cumulative optimal resources. The algorithm guarantees that the best resources for cloud processing are selected by the overall system. The system stays efficient by allocating resources that will provide the very best performance for cloud resources, while still guaranteeing functionality²⁷.

Step 4: smart learning output and real-time adaptation

Lastly, smart applications are used to leverage processed data into usable learning outputs. Adaptive content delivery systems that are learning resources specifically tailored to each individual student based on data compiled from AWS Lambda and profile storage help to create more specialized remediation for slow learners and advanced content for faster learners. Students will receive tailored feedback developed from real-time analytics.

Fig. 5

Processing and real-time visualization after detection. Labeling, class tagging, and confidence scoring are applied to unified detection findings, which are GPS/context-synchronized for action triggers and user interface presentation.

The actual visualizations of live performance, attendance, and behavioural trends in admin dashboards support teachers are shown in Fig. 5. The system might give messages or alerts (for example, disengagement or anomalies) to the administration or teachers, and they may intervene before something bad happens. Tablets, laptops, and internet portals give users access to the output through a defined end-user interface, which creates a consistent learning environment. This layer makes it possible to meet the goals of participation, feedback, and monitoring while keeping the successful and safe methods from earlier phases. It makes learning easy to get to, allows for continual learning, and lets us make changes quickly, even when resources are really limited.

Eq.n 10 establishes the strength of the system’s uptime and connectivity performance $\:Cp$ even with the unstable state $\:us$ of the network $\:Nt$.

$$\:Cp=us\left(Nt-{at}^{{\prime\:}{\prime\:}}\right)+ds\left({le}^{{\prime\:}}-nw\right)\text{*}ss\:\:$$

(10)

That is, the amount of time$\:{at}^{{\prime\:}{\prime\:}}$ the system is usable despite network outage/disruptions $\:ds,$ and Eq. 10 is for ensuring that learning experiences $\:{le}^{{\prime\:}}$remain uninterrupted or usable, even when the network is weak $\:nw$ or intermittent. This strength of the system $\:ss$ should allow it to support learners who are accessing content in more remote locations and from continuous-data-expensive networks without replacing the quality of their teaching and learning environment.

This Eq. 11 defines how anomalies $\:An$ in students’ behavior $\:s{b}^{{\prime\:}}$ (e.g., disengagement, low activity) are detected by comparing the baseline pattern $\:bl$ to real-time data.

$$\:An=s{b}^{{\prime\:}}\left(bl-is\right)=in\left(eb-{pr}^{{\prime\:}}\right)\text{*}\left[{rp}^{{\prime\:}}-al\right]$$

(11)

The system can provide alerts to instructors $\:is$ or can automatically suggest interventions $\:in$ based on deviations from expected behavior $\:eb$. The Eq. 11 serves to keep students engaged while the system is being predictive $\:{pr}^{{\prime\:}}$and responsive $\:{rp}^{{\prime\:}}$to anything interfering with the ability to learn $\:al$, to keep a proactive approach to student support²⁸.

This Eq. 12 represents the system’s capability $\:Sc$ of remaining operational and online performance under imperfect network conditions, where poor network conditions can mean the system remains up for various amounts of time and down or disrupted at other times.

$$\:{Sc=oR}_{xc}-fc\left(le-p{r}^{{\prime\:}}\right)+ii\text{*}rl\left(ol-b{l}^{{\prime\:}}{\prime\:}\right)\:\:$$

(12)

Eq.12 involves the actual operational values $\:{Sc=oR}_{xc}$and is for the purpose of facilitating continuity $\:fc$ in a learning experience and movement $\:\left(le-p{r}^{{\prime\:}}\right)$to the learning in a poor often intermittent network environment $\:ii$. This resilience $\:rl$ allows the system to operate and allow learning $\:ol$ with learners who may be in remote or bandwidth-limited $\:b{l}^{{\prime\:}}{\prime\:}$situations while ensuring their education is not compromised.

Fig. 6

End-to-end workflow for secure, adaptive smart learning. IoT data collection is followed by edge preprocessing, secure AWS cloud analytics, real-time output, and continuous feedback to optimize learning processes.

The first part of the SLICED framework is data capture in the context of the learning environment, from IoT devices, including sensors and wearables, as shown in Fig. 6. The edge layer does preprocessing and filtering with the data and other processes to enhance relevancy and improve latency. The data is then transmitted through a secure channel through AWS IoT Core. Data is encrypted while it is in transit, processed in real-time, and stored in a secure environment in AWS. Smart learning applications provide features like analytics dashboards, real-time feedback, and custom content delivery based on the information collected while the student and teacher end users are interacting through simple-to-use, connected devices.

The SLICED platform makes learning safe by using a structured data flow. It does this by gathering data from IoT, filtering it at the edge, processing it in the cloud, and delivering material that changes based on what the system might need. It makes sure that communication is safe, learning feedback is smart, and responses happen in real time. By combining AWS Cloud with the Internet of Things, the system creates a dynamic, safe learning environment that is in line with educational goals. This helps the system get around problems it had before.

Implementation details

Dataset description

The “Smart Classroom IoT-Edge Dataset” on Kaggle is a simulated, real-time dataset representative of typical smart classroom environments powered by IoT and edge computing. In this study, the dataset is primarily used for simulation and validation purposes to evaluate the performance of the proposed SLICED framework. It provides multimodal interaction data generated by IoT sensors, mirroring real-world dynamics for behavior analysis, adaptive learning response, and environmental monitoring. This enables robust testing of anomaly detection models, resource allocation algorithms, and personalized content delivery mechanisms under practical, realistic conditions. The dataset supports benchmarking against existing methods by providing a standardized input for system response, security, and scalability evaluations, helping validate the effectiveness of SLICED in enabling secure, low-latency, and adaptive learning experiences²⁹.

Tech stack

• AWS IoT Core – Enables secure device connectivity and communication between edge devices and the cloud.

• AWS Lambda – Provides serverless compute capabilities for real-time data processing and automation.

• AWS KMS (Key Management Service) – Ensures encryption and secure key management for protecting sensitive student data.

• Edge Devices (IoT-enabled sensors and hardware) – Used for real-time data collection, local processing, and initial filtering to reduce latency.

• Cloud Infrastructure (AWS Cloud) – Supports scalable storage, processing, and orchestration of learning resources.

• Adaptive Cloud–Edge Integration – Coordinates dynamic resource allocation and real-time responses to ensure uninterrupted, intelligent learning experiences.

AWS IoT Core was chosen for its strong device authentication and end-to-end encrypted connectivity, assuring data privacy and dependability from edge devices to the cloud. AWS Lambda enables real-time, serverless event processing, enabling the framework to automate adaptive learning actions instantly and scale without server maintenance. Industry-standard encryption and centralized key management protect sensitive educational data across remote resources with AWS KMS. Real-time classroom responsiveness requires early data filtering and latency reduction by edge devices. Cloud infrastructure provides seamless orchestration, scalable storage, and adaptive cloud–edge interaction for dynamic resource allocation and context-aware learning under varying loads.

Baseline models for comparison

• Traditional Cloud-Centric Learning Systems (centralized processing without edge filtering).

• Standard IoT-Based Architectures (basic sensor-to-cloud setups without adaptive resource management).

• Conventional Learning Management Systems (LMS) (without integrated cloud–edge security or real-time adaptability).

• Basic Encryption Models (standalone AES or RSA without AWS KMS orchestration).

• Edge-Only Computing Models (without cloud-based automation and scalability).

• Cloud-Only Processing Pipelines (lacking local edge preprocessing and latency reduction).

Simulation setup

The simulation used 50–100 edge nodes (IoT-enabled sensors) to represent student and instructor devices across several locations, simulating a real-world deployment. Traditional Wi-Fi (802.11ac) and simulated 4G cellular networks were used to measure latency and connection. System stability and peak-load behavior were tested over 24 h in each scenario. IoT Core controlled device connectivity, Lambda processed real-time events, and KMS safeguarded critical data transmission. DynamoDB and S3 housed user data and instructional materials, respectively, with CloudWatch monitoring resource utilization and system performance. User logins, frequent content access, and real-time streaming were system load criteria, with 500 simultaneous requests at peak. The simulation used Raspberry Pi 4 devices, AWS IoT Device Simulator virtual IoT nodes, and laptops to simulate classroom device diversity and improve reproducibility.

Using a variety of competing approaches (SL-IoT, LMS-IoT, PEF-En, R-BEC, H-Ed-AI, Cc-ELn, Azure IoT, Google Cloud IoT, and SLICED), the experimental protocol assesses four essential components of the Internet of Things–edge–cloud innovative learning architecture. Replaying similar workloads with up to 500 concurrent queries and determining the average end-to-end delay in comparison to a baseline for a non-Internet of Things learning management system (LMS) is how latency reduction is quantified. Data security evaluation is conducted using programmed attack scenarios, which yield a breach-attempt percentage. Lower numbers imply a higher level of protection. When aiming for edge filtering accuracy, it is necessary to inject labelled streams of relevant and noisy events at the edge, then calculate the fraction of data correctly maintained and the fraction eliminated. The strength of user authentication may be described as a composite score that combines attributes such as multi-factor authentication, token policies, and resistance to brute-force and credential-stuffing attacks. When taken as a whole, these studies provide a replicable, measurable foundation for evaluating and contrasting latency, security, edge intelligence, and access control across all implemented approaches.

Results

The eight measures of latency, data security, resource scaling, edge filtering accuracy, system response time, disruption resilience, automated processes, and good user authentication create a unique SLICED framework in this study. This would give us a view of the scalability, reliability, and overall performance of the SLICED framework idea when applied in real-world classrooms, when using AWS, Cloud, IoT, edge technologies to optimize reliable, safe online education.

Analysis of latency reduction

Fig. 7

Latency reduction analysis for various IoT-enabled smart learning frameworks. SLICED demonstrates the lowest latency among compared methods, highlighting its effectiveness in minimizing end-to-end processing delays.

SLICED utilizes AWS Lambda and IoT Core to process data at the edge and help significantly reduce latency, creating ideal conditions for learners to receive immediate feedback and to keep their experience uninterrupted, as analyzed in Fig. 7. For many learners, especially in parts of the world with slow internet access or learners who may not be near cities, this is incredibly beneficial, using Eq. 13. As a corollary, 27% latency reduction was realized by the system, which improved responsiveness in live learning sessions and reduced the average time taken to execute a task to 248 m/s from the previous 340 m/s.

The below Eq. 13 measures the latency reduction $\:Lr$ of the SLICED system over standard systems, by assessing the average $\:{aa}^{{\prime\:}}$difference in the time required $\:tr$ to acquire data and the time required to process data.

$$\:Lr=\left({aa}^{{\prime\:}}-tr\right)+qltr\left(rt-o{t}^{{\prime\:}}\right)\text{*}dt\left(E{t}^{{\prime\:}}-pr\right)$$

(13)

The goal is to quantify latency reduction $\:qltr$ in Eq. 13, which makes it possible to go beyond latency reduction into the real-time $\:rt$ optimality of the SLICED system $\:qltr\left(rt-o{t}^{{\prime\:}}\right)$by focusing on how data transmission $\:dt$ and processing time $\:dt\left(E{t}^{{\prime\:}}-pr\right)$can be minimized.

Analysis of data security

Fig. 8

Data security analysis based on breach attempt percentages across smart learning methods. SLICED shows the lowest breach attempt rate, indicating superior data security compared to existing frameworks and major IoT platforms.

SLICED utilizes AWS KMS for encryption and communicates through secure IoT Core channels as a means of keeping student data safe when in transit and storage, as illustrated in Fig. 8. To provide further protection to the system, real-time authentication adds another layer of security. The purpose of these security features is to minimize the chance of a data leak. Results indicated a 33% improvement in data security metrics and a 61% reduction in attempted unauthorized access by Eq. 14 when compared to baseline cloud learning systems.

This Eq. 14 determines the security score $\:ss$ for data transfer across the SLICED system.

$$\:ss=dt\left(ef-sm\right)+\left({au}^{{\prime\:}}-lba\left(hd-p{t}^{{\prime\:}{\prime\:}}\right)\right)$$

(14)

The above Eq. 14 calculates the breach attempts out of total data transactions $\:dt$ and can show some effectiveness $\:ef$ of security mechanisms $\:sm$ including encryption and authentication$\:{au}^{{\prime\:}}$, i.e., a lower breach attempts $\:lba$ percentage indicates higher data security $\:hd$, which highlights SLICED’s encryption policies$\:\left({au}^{{\prime\:}}-lba\left(hd-p{t}^{{\prime\:}{\prime\:}}\right)\right)$, and secure cloud processing mechanisms for data security.

Analysis of resource scalability

Table 2 Comparing smart learning with IoT resource scalability. With 90% scalability, dynamic cloud-edge resource management makes SLICED the best choice for flexible and large-scale deployments.

SLICED leverages demand-based resource allocation and scalability while managing AWS cloud and edge resources are examined in Table 2. With the serverless structure and seamless integration of Lambda triggers, the platform is able to allocate resources based on traffic, optimizing resource utilization during peak traffic periods using Eq. 15. This flexibility allowed for an overall better cloud resource utilization of 40%, with idle resource time decreased by 25%, which guarantees cost-effectiveness and ensures system redundancy across different user loads.

Eq. 15 measures how scalable a resource $\:Scr$ is by comparing the resources allocated $\:ra$ to the resources actually used.

$$\:Scr=ra\left({ef}^{{\prime\:}}-{ru}^{{\prime\:}{\prime\:}}\right)+\left(ls-gf{a}^{{\prime\:}{\prime\:}}\right)\text{*}\left(sc-vg\right)$$

(15)

A high score indicates efficient $\:{ef}^{{\prime\:}}$usage of resource$\:{ru}^{{\prime\:}{\prime\:}}$, while a low score $\:ls$ may indicate that the resource(s) are over-provisioned, as mentioned in Eq. 15. For SLICED, this equation is used to assess whether the system is capable of scaling dynamically$\:\left(sc-vg\right)$ to meet usage demand (from low to high loading) while minimising the use of cloud resources, because managing cloud resources is an important aspect of scalability $\:\left(ls-gf{a}^{{\prime\:}{\prime\:}}\right)$to a growing education system.

Analysis of edge filtering accuracy

SLICED applies edge filtering to discard redundant data prior to coming to the cloud, leading to better analytics and better decisions in Fig. 9. This is how effective edge data processing is accomplished with SLICED. SLICED improved data accuracy up to 29% in tests, especially in noisy situations with a variety of connected sensors, giving educators and school administrators more confidence in their decisions by Eq. 16.

This Eq. 16 determines the accuracy $\:ac$ of edge filtering $\:ed$, assessing the processed (filtered) data in relation to the raw data inputs.

$$\:ac=ed\left(me-rd\right)+fl\left({sd}^{{\prime\:}}-{ca}^{{\prime\:}{\prime\:}}\right)\text{*}ab-{cb}^{{\prime\:}}\:\:\:$$

(16)

The purpose of Eq. 16 is to measure the efficacy $\:me$ of irrelevant or redundant data $\:rd$ filtration at the edge level $\:fl$, ensuring that the most salient data $\:{sd}^{{\prime\:}}$are sent to the cloud for analysis$\:{ca}^{{\prime\:}{\prime\:}}$. This gives the system, both cloud and edge $\:ab$, the ability to conserve bandwidth and processing capabilities $\:{cb}^{{\prime\:}}$.

Fig. 9

Edge filtering accuracy across smart learning and IoT architectures. SLICED achieves the highest filtering accuracy, demonstrating superior data preprocessing performance compared to contemporary educational and cloud IoT solutions.

Analysis of system response time

Table 3 IoT vs. smart learning system reaction time. Efficient edge processing gives SLICED the quickest system response at 150 ms, beating cloud-centric and standard IoT frameworks.

AWS Lambda implements event-driven architecture which SLICED utilizes for fast execution of student engagements, whether submitting quizzes or accessing content; as a result, wait times were halved shown in Table 3. when matched against centralized architectures, the system examines a 20 desires need 20% improvement response time, jeb.js Ajax polling average time for query processing Puente a range of306 with average query funding with Eq. 17. Auto completion 312ms to 250ms, which allows for more directed and smoother digital learning.

This Eq. 17 quantifies system response time$\:{srt}^{{\prime\:}}$; it measures time from the moment the input data $\:{i}_{p}$is acquired when the response is produced by the edge nodes$\:{en}^{{\prime\:}}$.

$$\:{srt}^{{\prime\:}}=\left({i}_{p}\left({hq}^{{\prime\:}}-rt\right)\right)+{cd}^{{\prime\:}}\left({S}_{e}f-{rt}^{{\prime\:}}\right)\text{*}O{z}^{{\prime\:}{\prime\:}}$$

(17)

A low response time indicates that the system is of high quality$\:{hq}^{{\prime\:}}$ in terms of SLICED, the system want to minimize response time such that real-time learning interactions$\:\left({i}_{p}\left({hq}^{{\prime\:}}-rt\right)\right)$ by Eq. 17, such as adaptive content delivery $\:{cd}^{{\prime\:}}$or student feedback$\:{S}_{e}f$, happen in real time$\:{rt}^{{\prime\:}}$, and optimize $\:O{z}^{{\prime\:}{\prime\:}}$ a better user experience and system efficiency.

Analysis of connectivity resilience

Table 4 Evaluation of smart learning framework connectivity resilience. Compare uptime percentages and reconnection delays, and SLICED has the strongest operational resilience and fastest recovery, exceeding IoT and cloud-based techniques.

SLICED uses offline caching and edge processing to maintain learning even in settings with interrupted connectivity is analysed in Table 4. All data will sync when connection is re-established. However, during in-class monitoring, this method kept the system available 96% of the time (even only using simulated low bandwidth), and automatically recovered in less than 2.5 s using Eq. 18.

Eq. 18 provides a metric for connectivity resilience $\:Cr$, which indicates how well our system is still functioning in the face of changes to the network conditions $\:NC$ (stable, weak, or interrupted).

$$\:{Cr=}_{NC}\:\sum\:_{\left|\left|hr\right|\right|}^{2}{\int\:}_{i}^{n}\left|\left|ai\left({\forall\:}_{cf}-{\partial\:}_{ro}\right)\right|\right|-\left|\left|{nu}\left|{T}_{o}+{C}_{s}\right|\right|\right|\:$$

(18)

A high resilience score indicates that some resilience$\:|\left|hr\right|$ is constructed into the system that allows it to absorb interruptions $\:ai$ using Eq. 18 and continue functioning$\:{\forall\:}_{cf}$, meaning that, for example, the SLICED platform can remain operational$\:{\partial\:}_{ro}$ while the network is unstable $\:nu$, rather than terminating operationally$\:{T}_{o}$. The SLICED platform must allow the learner to maintain consistent access$\:{C}_{s}$ to learning content, rather than an interruption $\:\left|\left|{nu}\left|{T}_{o}+{C}_{s}\right|\right|\right|$in access to content, causing disruption in learning.

Analysis of automation efficiency

Table 5 Automation efficiency and impact analysis of various IoT-enabled smart learning methods. The table summarizes time saved and error rates, showing SLICED offers the greatest automation efficiency and lowest errors among all compared frameworks.

SLICED automates many processes with AWS Lambda, including logging, data backup, and issue creation. Therefore, it enables the platform to develop and respond to end-user needs quicker, with less human disruption shown in Table 5. With automation, the framework decreased human error by 45% and decreased time processing data manually by 37% by Eq. 19 and this made operations smoother for the administrators and the instructors.

Eq.19 quantifies the efficiency obtained from automation by comparing the time on manual work with the time saved with automation$\:\left(Ca\right)$.

$$\:\left(Ca\right)={t}_{s}-ie+\:{c}_{f},\:\:\:\:\left(dp\times\:cp\right)+\:{A}_{p-1}\left(dc\right)$$

(19)

This is intended to show time saved $\:{t}_{s}$and improved efficiency $\:ie$ with automation to cloud functions $\:{c}_{f}$such as data processing $\:dp$ and content personalization $\:cp$ by Eq. 19. In the case of SLICED, automation$\:{A}_{p-1}$ provides timeliness and more accurate decisions$\:\left(dc\right)$, improved user experience and impact of the system overall.

Analysis of user authentication strength

Fig. 10

User authentication strength analysis for different IoT-based frameworks. SLICED demonstrates the highest authentication strength, highlighting its enhanced capability for secure access control in smart learning systems.

The SLICED user verification process implements multi-factor authentication within AWS Identity and IoT regulations, which in turn prevents impersonation and unauthorized access. There will be secure access for both students and staff due to this process, illustrated in Fig. 10. The SLICED authentication model surpassed baseline models, which showed an average of 81% accuracy in creating a valid user authentication for students and staff, as there was a demonstrated authentication success rate of 98.6% in evaluations by Eq. 20, showing further verification that the SLICED platform satisfied the integrity and compliance measures of secure learning.

Eq. 20 measures the strength of the user authentication mechanism $\:U\left(ai\right)$.

$$\:U\left(ai\right)=\:\sum\:_{i=1}^{n}{S}_{a}\left(at\right)\times\:{h}^{r}-{a}^{p}\left(ua\left(pr-sc\right)-nat\right)$$

(20)

More specifically, it measures the percentage of successful authentications$\:{S}_{a}\left(at\right)$, A where the higher result $\:{h}^{r}$indicates a better authentication process$\:{a}^{p}$ since this means fewer unauthorized access $\:ua$ attempts through utilization of equtaion 20. This is important in the SLICED system because privacy $\:pr$ and security $\:sc$ are very important for the individuals whose data are contributed. This metric helps the system ensure that non-authorized users $\:nat$ are not able to access protected learning materials and personalized learning content. Eight metric variations of performance improvement included a 27% reduction of latency, a 33% enhancement of security, and a 20% boost of responsiveness; collaboration further improved by automation, authentication, and filtering accuracy; cloud edge collaboration provided resiliency and coordinated scalability of resources. Collectively, these confirmations substantiate the conclusion that SLICED effectively provides a contemporary educational transformation with a safe, efficient, and flexible learning systems infrastructure.

Discussion

This paper presents the proposed system, SLICED, as a practical, scalable, and secure solution for digital learning spaces. This study uses simulated environments rather than live deployment of SLICED, which has 27% lower latency, 33% better data security, and 20% faster response than centralized solutions. AWS’s unique infrastructure, anticipated operational expenses, and limited generalizability without field testing are major restrictions. Real-world deployment pilots across varied schools will test the platform’s scalability, cost-effectiveness, and robustness under varying network conditions. SLICED will add AI-enabled analytics for real-time personalization, systematic blockchain-based security testing for distributed classrooms, and multilingual support. The platform will also test privacy-preserving and federated learning models in stringent security environments to address data governance and global educational compliance. Integrating AI-enabled analytics for personalized learning and blockchain-enabled distributed ledgers for infrastructure security may enhance the SLICED experience. The proposed system could make learning more immersive by utilizing augmented or virtual reality technologies. Another benefit to consider is providing multiple languages to learners at once to diversify access for potential users of the SLICED platform. From there, the system could assess how successful the SLICED process is and how scalable it is in other learning contexts, such as schools in developing countries or large university communities. The system could even examine more rigid security environments in addition to privacy-preserving models and federated learning. With the improvements above, SLICED may become a pivotal component in smart education.

The SLICED design revealed statistically significant advantages compared to all baselines. By achieving the lowest mean latency (about 242 milliseconds, with a standard deviation of approximately 9 milliseconds), it outperformed both the conventional SL-IoT (approximately 320 milliseconds) and the major cloud platforms, such as Google Cloud IoT (approximately 249 milliseconds). The accuracy of edge filtering achieved around 96%, which is significantly higher than the 85–92% achievable by competing approaches. This indicates that redundant data was removed with greater precision before cloud upload. The results of the security tests demonstrated that SLICED prevented approximately 97% of programmed breaches and achieved the highest composite authentication-strength score (approximately 21/25). This substantiates that its performance enhancements do not compromise confidentiality or access control.

Limitations

Evaluation context and deployment assumptions are fundamental SLICED restrictions. First, the gains in latency, security, and responsiveness may not apply to all school infrastructures and user behaviors, as the findings are from controlled simulations rather than long-term production rollouts. Second, AWS reliance restricts mobility, cost modeling, and application in legislative or procurement-restricted areas. Third, operational costs, edge-device heterogeneity, and network unpredictability in low-resource environments are poorly understood. Advanced extensions (AI analytics, federated learning, blockchain, AR/VR, multilingual support) are planned yet untested.