OnDev-LCT: On-Device Lightweight Convolutional Transformers towards federated learning.

Federated learning (FL) has emerged as a promising approach to collaboratively train machine learning models across multiple edge devices while preserving privacy. The success of FL hinges on the efficiency of participating models and their ability to handle the unique challenges of distributed learning. While several variants of Vision Transformer (ViT) have shown great potential as alternatives to modern convolutional neural networks (CNNs) for centralized training, the unprecedented size and higher computational demands hinder their deployment on resource-constrained edge devices, challenging their widespread application in FL. Since client devices in FL typically have limited computing resources and communication bandwidth, models intended for such devices must strike a balance between model size, computational efficiency, and the ability to adapt to the diverse and non-IID data distributions encountered in FL. To address these challenges, we propose OnDev-LCT: Lightweight Convolutional Transformers for On-Device vision tasks with limited training data and resources. Our models incorporate image-specific inductive biases through the LCT tokenizer by leveraging efficient depthwise separable convolutions in residual linear bottleneck blocks to extract local features, while the multi-head self-attention (MHSA) mechanism in the LCT encoder implicitly facilitates capturing global representations of images. Extensive experiments on benchmark image datasets indicate that our models outperform existing lightweight vision models while having fewer parameters and lower computational demands, making them suitable for FL scenarios with data heterogeneity and communication bottlenecks.

Contrastive encoder pre-training-based clustered federated learning for heterogeneous data.

Federated learning (FL) is a promising approach that enables distributed clients to collaboratively train a global model while preserving their data privacy. However, FL often suffers from data heterogeneity problems, which can significantly affect its performance. To address this, clustered federated learning (CFL) has been proposed to construct personalized models for different client clusters. One effective client clustering strategy is to allow clients to choose their own local models from a model pool based on their performance. However, without pre-trained model parameters, such a strategy is prone to clustering failure, in which all clients choose the same model. Unfortunately, collecting a large amount of labeled data for pre-training can be costly and impractical in distributed environments. To overcome this challenge, we leverage self-supervised contrastive learning to exploit unlabeled data for the pre-training of FL systems. Together, self-supervised pre-training and client clustering can be crucial components for tackling the data heterogeneity issues of FL. Leveraging these two crucial strategies, we propose contrastive pre-training-based clustered federated learning (CP-CFL) to improve the model convergence and overall performance of FL systems. In this work, we demonstrate the effectiveness of CP-CFL through extensive experiments in heterogeneous FL settings, and present various interesting observations.

Self-Organizing Democratized Learning: Toward Large-Scale Distributed Learning Systems.

Emerging cross-device artificial intelligence (AI) applications require a transition from conventional centralized learning systems toward large-scale distributed AI systems that can collaboratively perform complex learning tasks. In this regard, democratized learning (Dem-AI) lays out a holistic philosophy with underlying principles for building large-scale distributed and democratized machine learning systems. The outlined principles are meant to study a generalization in distributed learning systems that go beyond existing mechanisms such as federated learning (FL). Moreover, such learning systems rely on hierarchical self-organization of well-connected distributed learning agents who have limited and highly personalized data and can evolve and regulate themselves based on the underlying duality of specialized and generalized processes. Inspired by Dem-AI philosophy, a novel distributed learning approach is proposed in this article. The approach consists of a self-organizing hierarchical structuring mechanism based on agglomerative clustering, hierarchical generalization, and corresponding learning mechanism. Subsequently, hierarchical generalized learning problems in recursive forms are formulated and shown to be approximately solved using the solutions of distributed personalized learning problems and hierarchical update mechanisms. To that end, a distributed learning algorithm, namely DemLearn, is proposed. Extensive experiments on benchmark MNIST, Fashion-MNIST, FE-MNIST, and CIFAR-10 datasets show that the proposed algorithm demonstrates better results in the generalization performance of learning models in agents compared to the conventional FL algorithms. The detailed analysis provides useful observations to further handle both the generalization and specialization performance of the learning models in Dem-AI systems.

Clouds Proportionate Medical Data Stream Analytics for Internet of Things-Based Healthcare Systems.

Internet of Things (IoT) assisted healthcare systems are designed for providing ubiquitous access and recommendations for personal and distributed electronic health services. The heterogeneous IoT platform assists healthcare services with reliable data management through dedicated computing devices. Healthcare services' reliability depends upon the efficient handling of heterogeneous data streams due to variations and errors. A Proportionate Data Analytics (PDA) for heterogeneous healthcare data stream processing is introduced in this manuscript. This analytics method differentiates the data streams based on variations and errors for satisfying the service responses. The classification is streamlined using linear regression for segregating errors from the variations in different time intervals. The time intervals are differentiated recurrently after detecting errors in the stream's variation. This process of differentiation and classification retains a high response ratio for healthcare services through spontaneous regressions. The proposed method's performance is analyzed using the metrics accuracy, identification ratio, delivery, variation factor, and processing time.

Internet of things for secure surveillance for sewage wastewater treatment systems.

IoT is a secure communication technology used to transfer data from a physical entity to a device with intelligent analysis tools through a wireless channel. The wastewater treatment method extracts pollutants and transforms them into effluents added to the water supply with minimal environmental effects or recovered directly. The major issue is monitoring the disposal of sewage in the treatment plants. Hence, this paper, Surveillance-based Sewage Wastewater Monitoring System (SSWMS) with IoT, has been proposed for monitoring wastewater treatment and improving water quality. A smart water sensor enabled by IoT monitors water quality, water pressure, and water temperature and quantifies water dynamics to map water flow through the entire treatment facility. The proposed method calculates the wastewater treatment facility's effectiveness and ensures that chemical releases are maintained below allowable levels. Thus, the experimental results show the improved recycling water quality level is raised to 97.98%, enhancing secure communication and less moisture content when compared to other methods.

When CVaR Meets With Bluetooth PAN: A Physical Distancing System for COVID-19 Proactive Safety.

In this work, we propose a risk-aware physical distancing system to assure a private safety distance from others for reducing the chance of being affected by the COVID-19 or such kind of pandemic. In particular, we have formulated a physical distancing problem by capturing Conditional Value-at-Risk (CVaR) of a Bluetooth-enabled personal area network (PAN). To solve the formulated risk-aware physical distancing problem, we propose two stages solution approach by imposing control flow, linear model, and curve-fitting schemes. Notably, in the first stage, we determine a PAN creator's safe movement distance by proposing a probabilistic linear model. This scheme can effectively cope with a tail-risk from the probability distribution by satisfying the CVaR constraint for estimating safe movement distance. In the second stage, we design a Levenberg-Marquardt (LM)-based curve fitting algorithm upon the recommended safety distance and current distances between the PAN creator and others to find an optimal high-risk trajectory plan for the PAN creator. Finally, we have performed an extensive performance analysis using state-of-the-art Bluetooth data to establish the proposed risk-aware physical distancing system's effectiveness. Our experimental results show that the proposed solution approach can effectively reduce the risk of recommending safety distance towards ensuring private safety. In particular, for a 95% CVaR confidence, we can successfully deal with 45.11% of the risk for measuring the PAN creator's safe movement distance.

Controlling the Outbreak of COVID-19: A Noncooperative Game Perspective.

COVID-19 is a global epidemic. Till now, there is no remedy for this epidemic. However, isolation and social distancing are seemed to be effective preventive measures to control this pandemic. Therefore, in this article, an optimization problem is formulated that accommodates both isolation and social distancing features of the individuals. To promote social distancing, we solve the formulated problem by applying a noncooperative game that can provide an incentive for maintaining social distancing to prevent the spread of COVID-19. Furthermore, the sustainability of the lockdown policy is interpreted with the help of our proposed game-theoretic incentive model for maintaining social distancing where there exists a Nash equilibrium. Finally, we perform an extensive numerical analysis that shows the effectiveness of the proposed approach in terms of achieving the desired social-distancing to prevent the outbreak of the COVID-19 in a noncooperative environment. Numerical results show that the individual incentive increases more than 85% with an increasing percentage of home isolation from 25% to 100% for all considered scenarios. The numerical results also demonstrate that in a particular percentage of home isolation, the individual incentive decreases with an increasing number of individuals.

Flexible computation offloading in a fuzzy-based mobile edge orchestrator for IoT applications.

In the Internet of Things (IoT) era, the capacity-limited Internet and uncontrollable service delays for various new applications, such as video streaming analysis and augmented reality, are challenges. Cloud computing systems, also known as a solution that offloads energy-consuming computation of IoT applications to a cloud server, cannot meet the delay-sensitive and context-aware service requirements. To address this issue, an edge computing system provides timely and context-aware services by bringing the computations and storage closer to the user. The dynamic flow of requests that can be efficiently processed is a significant challenge for edge and cloud computing systems. To improve the performance of IoT systems, the mobile edge orchestrator (MEO), which is an application placement controller, was designed by integrating end mobile devices with edge and cloud computing systems. In this paper, we propose a flexible computation offloading method in a fuzzy-based MEO for IoT applications in order to improve the efficiency in computational resource management. Considering the network, computation resources, and task requirements, a fuzzy-based MEO allows edge workload orchestration actions to decide whether to offload a mobile user to local edge, neighboring edge, or cloud servers. Additionally, increasing packet sizes will affect the failed-task ratio when the number of mobile devices increases. To reduce failed tasks because of transmission collisions and to improve service times for time-critical tasks, we define a new input crisp value, and a new output decision for a fuzzy-based MEO. Using the EdgeCloudSim simulator, we evaluate our proposal with four benchmark algorithms in augmented reality, healthcare, compute-intensive, and infotainment applications. Simulation results show that our proposal provides better results in terms of WLAN delay, service times, the number of failed tasks, and VM utilization.

DeepAuC: Joint deep learning and auction for congestion-aware caching in Named Data Networking.

Over the last few decades, the Internet has experienced tremendous growth in data traffic. This continuous growth due to the increase in the number of connected devices and platforms has dramatically boosted content consumption. However, retrieving content from the servers of Content Providers (CPs) can increase network traffic and incur high network delay and congestion. To address these challenges, we propose a joint deep learning and auction-based approach for congestion-aware caching in Named Data Networking (NDN), which aims to prevent congestion and minimize the content downloading delays. First, using recorded network traffic data on the Internet Service Provider (ISP) network, we propose a deep learning model to predict future traffic over transit links. Second, to prevent congestion and avoid high latency on transit links, which may experience congestion in the future; we propose a caching model that helps the ISP to cache content that has a high predicted future demand. Paid-content requires payment to be downloaded and cached. Therefore, we propose an auction mechanism to obtain paid-content at an optimal price. The simulation results show that our proposal prevents congestion and increases the profits of both ISPs and CPs.

A Novel Maximum Entropy Markov Model for Human Facial Expression Recognition.

Research in video based FER systems has exploded in the past decade. However, most of the previous methods work well when they are trained and tested on the same dataset. Illumination settings, image resolution, camera angle, and physical characteristics of the people differ from one dataset to another. Considering a single dataset keeps the variance, which results from differences, to a minimum. Having a robust FER system, which can work across several datasets, is thus highly desirable. The aim of this work is to design, implement, and validate such a system using different datasets. In this regard, the major contribution is made at the recognition module which uses the maximum entropy Markov model (MEMM) for expression recognition. In this model, the states of the human expressions are modeled as the states of an MEMM, by considering the video-sensor observations as the observations of MEMM. A modified Viterbi is utilized to generate the most probable expression state sequence based on such observations. Lastly, an algorithm is designed which predicts the expression state from the generated state sequence. Performance is compared against several existing state-of-the-art FER systems on six publicly available datasets. A weighted average accuracy of 97% is achieved across all datasets.

Web of Objects Based Ambient Assisted Living Framework for Emergency Psychiatric State Prediction.

Ambient assisted living can facilitate optimum health and wellness by aiding physical, mental and social well-being. In this paper, patients' psychiatric symptoms are collected through lightweight biosensors and web-based psychiatric screening scales in a smart home environment and then analyzed through machine learning algorithms to provide ambient intelligence in a psychiatric emergency. The psychiatric states are modeled through a Hidden Markov Model (HMM), and the model parameters are estimated using a Viterbi path counting and scalable Stochastic Variational Inference (SVI)-based training algorithm. The most likely psychiatric state sequence of the corresponding observation sequence is determined, and an emergency psychiatric state is predicted through the proposed algorithm. Moreover, to enable personalized psychiatric emergency care, a service a web of objects-based framework is proposed for a smart-home environment. In this framework, the biosensor observations and the psychiatric rating scales are objectified and virtualized in the web space. Then, the web of objects of sensor observations and psychiatric rating scores are used to assess the dweller's mental health status and to predict an emergency psychiatric state. The proposed psychiatric state prediction algorithm reported 83.03 percent prediction accuracy in an empirical performance study.

Critical Factors Influencing Decision to Adopt Human Resource Information System (HRIS) in Hospitals.

The aim of this research is to explore factors influencing the management decisions to adopt human resource information system (HRIS) in the hospital industry of Bangladesh-an emerging developing country. To understand this issue, this paper integrates two prominent adoption theories-Human-Organization-Technology fit (HOT-fit) model and Technology-Organization-Environment (TOE) framework. Thirteen factors under four dimensions were investigated to explore their influence on HRIS adoption decisions in hospitals. Employing non-probability sampling method, a total of 550 copies of structured questionnaires were distributed among HR executives of 92 private hospitals in Bangladesh. Among the respondents, usable questionnaires were 383 that suggesting a valid response rate of 69.63%. We classify the sample into 3 core groups based on the HRIS initial implementation, namely adopters, prospectors, and laggards. The obtained results specify 5 most critical factors i.e. IT infrastructure, top management support, IT capabilities of staff, perceived cost, and competitive pressure. Moreover, the most significant dimension is technological dimension followed by organisational, human, and environmental among the proposed 4 dimensions. Lastly, the study found existence of significant differences in all factors across different adopting groups. The study results also expose constructive proposals to researchers, hospitals, and the government to enhance the likelihood of adopting HRIS. The present study has important implications in understanding HRIS implementation in developing countries.

Human Behavior Analysis by Means of Multimodal Context Mining.

There is sufficient evidence proving the impact that negative lifestyle choices have on people's health and wellness. Changing unhealthy behaviours requires raising people's self-awareness and also providing healthcare experts with a thorough and continuous description of the user's conduct. Several monitoring techniques have been proposed in the past to track users' behaviour; however, these approaches are either subjective and prone to misreporting, such as questionnaires, or only focus on a specific component of context, such as activity counters. This work presents an innovative multimodal context mining framework to inspect and infer human behaviour in a more holistic fashion. The proposed approach extends beyond the state-of-the-art, since it not only explores a sole type of context, but also combines diverse levels of context in an integral manner. Namely, low-level contexts, including activities, emotions and locations, are identified from heterogeneous sensory data through machine learning techniques. Low-level contexts are combined using ontological mechanisms to derive a more abstract representation of the user's context, here referred to as high-level context. An initial implementation of the proposed framework supporting real-time context identification is also presented. The developed system is evaluated for various realistic scenarios making use of a novel multimodal context open dataset and data on-the-go, demonstrating prominent context-aware capabilities at both low and high levels.

An On-Demand Emergency Packet Transmission Scheme for Wireless Body Area Networks.

The rapid developments of sensor devices that can actively monitor human activities have given rise to a new field called wireless body area network (BAN). A BAN can manage devices in, on and around the human body. Major requirements of such a network are energy efficiency, long lifetime, low delay, security, etc. Traffic in a BAN can be scheduled (normal) or event-driven (emergency). Traditional media access control (MAC) protocols use duty cycling to improve performance. A sleep-wake up cycle is employed to save energy. However, this mechanism lacks features to handle emergency traffic in a prompt and immediate manner. To deliver an emergency packet, a node has to wait until the receiver is awake. It also suffers from overheads, such as idle listening, overhearing and control packet handshakes. An external radio-triggered wake up mechanism is proposed to handle prompt communication. It can reduce the overheads and improve the performance through an on-demand scheme. In this work, we present a simple-to-implement on-demand packet transmission scheme by taking into considerations the requirements of a BAN. The major concern is handling the event-based emergency traffic. The performance analysis of the proposed scheme is presented. The results showed significant improvements in the overall performance of a BAN compared to state-of-the-art protocols in terms of energy consumption, delay and lifetime.

mDurance: A Novel Mobile Health System to Support Trunk Endurance Assessment.

Low back pain is the most prevalent musculoskeletal condition. This disorder constitutes one of the most common causes of disability worldwide, and as a result, it has a severe socioeconomic impact. Endurance tests are normally considered in low back pain rehabilitation practice to assess the muscle status. However, traditional procedures to evaluate these tests suffer from practical limitations, which potentially lead to inaccurate diagnoses. The use of digital technologies is considered here to facilitate the task of the expert and to increase the reliability and interpretability of the endurance tests. This work presents mDurance, a novel mobile health system aimed at supporting specialists in the functional assessment of trunk endurance by using wearable and mobile devices. The system employs a wearable inertial sensor to track the patient trunk posture, while portable electromyography sensors are used to seamlessly measure the electrical activity produced by the trunk muscles. The information registered by the sensors is processed and managed by a mobile application that facilitates the expert's normal routine, while reducing the impact of human errors and expediting the analysis of the test results. In order to show the potential of the mDurance system, a case study has been conducted. The results of this study prove the reliability of mDurance and further demonstrate that practitioners are certainly interested in the regular use of a system of this nature.

A telemedicine application to schedule temperature in an in vivo sensor network for cancer treatment.

Wireless communication has played a significant role in modern healthcare systems. However, the death toll from chronic diseases, such as cancer, continues to increase. Hyperthermia combined with radiotherapy and/or chemotherapy is a promising strategy for cancer treatment, and temperature control is critical for the success of this intervention. In vivo sensors are an emerging technology in healthcare. Thermal awareness has also received attention in in vivo sensor research. In this context, we have been motivated to use in vivo sensors to regulate the temperature changes in cancer cells during combined treatment. Limitations in existing in vivo thermal-aware routing algorithms motivated us to use the in vivo "lightweight rendezvous routing" approach. However, smartphone-driven telemedicine applications are proliferating to provide remote healthcare and collaborative consultation, required in combined therapies. In this context, we have proposed a telemedicine application where a smartphone not only regulates temperature scheduling in in vivo sensors, but also communicates with local or remote clinicians to maintain collaborative efforts for combined therapies against cancer.

Data-centric multiobjective QoS-aware routing protocol for body sensor networks.

In this paper, we address Quality-of-Service (QoS)-aware routing issue for Body Sensor Networks (BSNs) in delay and reliability domains. We propose a data-centric multiobjective QoS-Aware routing protocol, called DMQoS, which facilitates the system to achieve customized QoS services for each traffic category differentiated according to the generated data types. It uses modular design architecture wherein different units operate in coordination to provide multiple QoS services. Their operation exploits geographic locations and QoS performance of the neighbor nodes and implements a localized hop-by-hop routing. Moreover, the protocol ensures (almost) a homogeneous energy dissipation rate for all routing nodes in the network through a multiobjective Lexicographic Optimization-based geographic forwarding. We have performed extensive simulations of the proposed protocol, and the results show that DMQoS has significant performance improvements over several state-of-the-art approaches.

ATLAS: a traffic load aware sensor MAC design for collaborative body area sensor networks.

In collaborative body sensor networks, namely wireless body area networks (WBANs), each of the physical sensor applications is used to collaboratively monitor the health status of the human body. The applications of WBANs comprise diverse and dynamic traffic loads such as very low-rate periodic monitoring (i.e., observation) data and high-rate traffic including event-triggered bursts. Therefore, in designing a medium access control (MAC) protocol for WBANs, energy conservation should be the primary concern during low-traffic periods, whereas a balance between satisfying high-throughput demand and efficient energy usage is necessary during high-traffic times. In this paper, we design a traffic load-aware innovative MAC solution for WBANs, called ATLAS. The design exploits the superframe structure of the IEEE 802.15.4 standard, and it adaptively uses the contention access period (CAP), contention free period (CFP) and inactive period (IP) of the superframe based on estimated traffic load, by applying a dynamic "wh" (whenever which is required) approach. Unlike earlier work, the proposed MAC design includes load estimation for network load-status awareness and a multi-hop communication pattern in order to prevent energy loss associated with long range transmission. Finally, ATLAS is evaluated through extensive simulations in ns-2 and the results demonstrate the effectiveness of the protocol.

MQ-MAC: a multi-constrained QoS-aware duty cycle MAC for heterogeneous traffic in wireless sensor networks.

Energy conservation is one of the striking research issues now-a-days for power constrained wireless sensor networks (WSNs) and hence, several duty-cycle based MAC protocols have been devised for WSNs in the last few years. However, assimilation of diverse applications with different QoS requirements (i.e., delay and reliability) within the same network also necessitates in devising a generic duty-cycle based MAC protocol that can achieve both the delay and reliability guarantee, termed as multi-constrained QoS, while preserving the energy efficiency. To address this, in this paper, we propose a Multi-constrained QoS-aware duty-cycle MAC for heterogeneous traffic in WSNs (MQ-MAC). MQ-MAC classifies the traffic based on their multi-constrained QoS demands. Through extensive simulation using ns-2 we evaluate the performance of MQ-MAC. MQ-MAC provides the desired delay and reliability guarantee according to the nature of the traffic classes as well as achieves energy efficiency.

Load-adaptive practical multi-channel communications in wireless sensor networks.

In recent years, a significant number of sensor node prototypes have been designed that provide communications in multiple channels. This multi-channel feature can be effectively exploited to increase the overall capacity and performance of wireless sensor networks (WSNs). In this paper, we present a multi-channel communications system for WSNs that is referred to as load-adaptive practical multi-channel communications (LPMC). LPMC estimates the active load of a channel at the sink since it has a more comprehensive view of the network behavior, and dynamically adds or removes channels based on the estimated load. LPMC updates the routing path to balance the loads of the channels. The nodes in a path use the same channel; therefore, they do not need to switch channels to receive or forward packets. LPMC has been evaluated through extensive simulations, and the results demonstrate that it can effectively increase the delivery ratio, network throughput, and channel utilization, and that it can decrease the end-to-end delay and energy consumption.