Federated learning (FL) was introduced to mitigate security risks in traditional machine learning (ML) and deep learning (DL), particularly concerning data privacy (DP), through its decentralized architecture. However, FL remains susceptible to cyber-security threats. Such risks arise when several legitimate client nodes are transformed into a zombie network. These compromised nodes can launch large-scale attacks that affect the confidentiality, integrity, and availability (CIA) of the FL network, ultimately leading to a decline in prediction metrics for legitimate clients. While previous studies have focused on specific, small-scale attacks, none have addressed the impact of large-scale attacks, specifically, an "attack coalition," where over 50% of the network is compromised, and multiple types of attacks occur simultaneously. Many existing studies rely on complex methods such as blockchain or malicious node detection, and only a few have considered the simpler alternative of "personalization," though with limitations. In this paper, we introduce the concept of attack coalition (AC), where malicious client nodes coordinate to undermine the FL system’s performance. We analyze the impact of this AC and propose a personalized FL approach to mitigate its effects. To validate our approach, we use a publicly available dataset measuring nurse stress levels, implemented through a deep neural network (DNN). Our results demonstrate that the AC severely reduces prediction metrics, with accuracy and F1-scores dropping to 18.34% and 18.68%, respectively. By applying our proposed method, these metrics improve significantly, reaching 99.36% and 98.46%, respectively.

Federated learning (FL) was introduced recently as a new machine learning (ML) paradigm. It is a distributed network of client nodes that train ML and deep learning (DL) models on their local data without sharing them to preserve data privacy (DP). However, these data are heterogeneous by nature as they are collected in different contexts using various sources such as IoT devices. Consequently, data heterogeneity (DH) in FL has brought new performance-related challenges. Few of these challenges have been addressed in the literature; moreover, context heterogeneity and balance rate were not explored at all. In this paper, we introduce an FL approach in which a trade-off between personalization and generalization is achieved to mitigate the impact of DH and obtain better performance. We focus on three DH challenges: context, non-independent and identically distributed (non-IID) data, and balance rate. For the implementation, fall detection (FD) data is used to demonstrate the potential of our approach in improving the FL system’s performance. FD is an important subject and is particularly prevalent for the safety of elderly people. Hence, we collected fall data from two sensors: accelerometer (ACC) and heart rate (HR), then, we used two ML models to evaluate our approach. We utilized XGBoost (XGB) for balanced and unbalanced clients and One-Class Support Vector Machine (OC-SVM) for one-label clients. Our approach achieved an average F1-score of 88%. A comparative study was also conducted with previous works on FD. Our results showed a performance improvement which exceeded 94.30% on average.

In recent years, significant advances in biometrics have essentially been driven by machine learning (ML) and deep learning (DL) progress. Numerous human identification applications are currently available using physical traits such as fingerprints, face, and voice. With the development of Internet of Things (IoT) sensors and the availability of a variety of ML algorithms, there has been increased research interest in subject identification (SI) based on behavioral cues. For example, several research works have been published on SI based on gait analysis. Sensors like accelerometers (ACC), gyroscopes, (GYR), and magnetometers (MAG) were used to collect data during limited activities such as walking. We believe that using data for one activity is not sufficient to adequately capture behavioral cues for the purpose of SI. Considering other cues such as gestures or head shaking and using a variety of sensors located on different parts of the human body are essential to developing a scenario that includes expressive human activities. We designed a specific scenario that included several activities, such as walking, giving a talk, chatting while sitting, and climbing stairs, using five inertial measurement units (IMU) located on various parts of the human body. Several ML algorithms, namely Linear Discriminant Analysis (LDA), K-Nearest Neighbours (KNN), Random Forest (RF), and XGBoost (XGB) were used. Our results show that SI yields beyond 99% accuracy for most activities. Furthermore, we succeeded in implementing a real-time IoT system for SI based on our best offline results. We achieved 98.04% accuracy within 0.06 ms of processing time (PT).

Fall is one of the main causes of injuries for the elderly, and fall detection (FD) for senior monitoring has received considerable attention from both the academic community and healthcare industries. In recent years, there has been an increasing interest in using wearable sensors, such as accelerometers to monitor the subject’s body movement and apply Machine Learning (ML) methods to detect and prevent falls. Since it is extremely difficult to collect accelerometer data of real falls during activities of daily living (ADL), researchers tended to rely on simulating falls in well-protected environments. They collected ADLs separately, applied ML algorithms to classify falls and ADLs, and reported very high FD accuracy rates. However, these studies cannot be applied in a real fall context. In this paper, instead of classifying ADL and fall separately, we propose to incorporate fall data within ADL data to obtain more realistic datasets and apply ML to detect falls. Several ML algorithms including CatBoost (CB), Decision Tree (DT), Random Forest (RF), and XGBoost (XGB) were applied to the datasets. Experimental results show a fall detection accuracy of $88.70 \%$. We also extend our work to cover slow fall which, to the best of our knowledge, was not extensively addressed in previous works.

Hand gesture recognition (HGR) is a vital area of research with widespread applications and employing disruptive technologies such as Artificial Intelligence (AI) and the Internet of Things (IoT). HGR is very important in human-computer interaction (HCI), especially for people with disabilities. Several approaches have been described in the literature, including image-based, radar-based, and haptic-based ones. However, these approaches are difficult to implement in real time as they require high processing time (PT). Inertial sensors (IS) offer a worthy alternative for HGR and require much less PT. Several research papers have been published in this domain. However, models capturing complex temporal patterns using traditional machine learning (ML) algorithms have limitations. In this paper, we propose a transformer-based approach to accurately recognize dynamic gestures. A dataset we collected from accelerometer (ACC) and gyroscope (GYR) sensors was used to evaluate the performance of our approach. Our experimental results achieved an accuracy of 96.77% and outperformed traditional ML algorithms in terms of accuracy using the same dataset: Linear Discriminant Analysis (LDA) with 58.06%, KNearest Neighbor (KNN) with 70.97%, XGBoost (XGB) with 74.19%, Convolutional neural network (CNN) with 74.19%, Random Forest (RF) with 77.42%, and Support vector machine (SVM) with 93.55%. Furthermore, we implemented a real-time HGR system that can achieve a very short response time of less than 300 ms.

Detecting Freezing of Gait (FOG) is crucial for Parkinson's disease (PD). Several research works have been published on FOG detection and prediction using machine learning (ML) and limited accelerometer data. FOG data collection is challenging and generally conducted on a limited number of persons as freezing occurs randomly, unlike walking which is a recurrent activity. This makes directly applying ML inefficient. In this paper, to improve previously published results, we apply three different ML algorithms: Linear Discriminant Analysis (LDA), Extreme Gradient Boosting (XGB), and Extra Trees (ET), in addition to a Transformer model to detect and predict FOG. This is achieved by modelling the publicly available DAPHNet dataset, which contains accelerometer readings from three wearable sensors. The data quality is first improved by using data augmentation, data balancing, and feature extraction. Our results show that accuracy (ACC), along with other metrics like sensitivity (SEN), specificity (SPE), and F1-score (F1) is important to quantify the performance of our results. From FOG detection results, we can see that XGB and ET algorithms provide 100% ACC, and slightly lower performance, 99.96% with LDA. For FOG prediction, our results prove we can achieve 99.21%, 96.40%, 96.21%, and 94.70% ACC with LDA, XGB, ET, and Transformer, respectively. These results outperform previously published results on the same dataset.

Falls are one of the most common causes of injury among the elderly. As a result, fall detection has received in the last decade considerable attention from both academia and the healthcare industry. Accelerometer data, collected from simulated falls, were widely used with classical machine learning (ML) algorithms as well as with threshold-based methods to identify fall situations that can be used to launch an alert for help. As collecting real fall data is challenging, most of the research papers on fall detection have used limited data which do not reflect the complexity of real fall situations. Fortunately, a comprehensive fall dataset called “Simulated Falls and Daily Living Activities Dataset” has recently become available. This dataset includes 1827 simulated falls of 20 different types. In this paper, we use this dataset to evaluate the possibility of fall detection, more precisely, impact and pre-impact which correspond to fall and pre-fall, respectively. Unlike the classical ML algorithms and threshold-based methods commonly used in previous research works, in this paper, we implement a bidirectional long short-term memory (Bi-LSTM) algorithm which we believe better reflects the impact and pre-impact context as it takes into consideration both backward and forward sequence information at every time step. Our experimental results showed that Bi-LSTM achieves an accuracy of 99.97% and 99.95%, with 99.80% and 99.30% sensitivity, and 100% and 99.99% specificity for fall and pre-fall detections, respectively. These results largely exceed previously published results.

This paper introduces a simple Wireless Sensor Network (WSN)-based framework that uses Proteus sensors of Libelium Smart Water Xtreme IoT platform to detect e-Coli in wastewater, uses an efficient priority-based routing protocol for timely notification of the detection e-Coli at the COVID-19 detection lab to identify the existence of SARS-CoV-2, the virus that currently causes the COVID-19 pandemic. These sensors use fluorescence to monitor coli forms in real-time, determining if the water is polluted and contaminated with SARS-CoV-2 once tested at the lab. The framework also includes an efficient Packet Priority Routing Protocol (PPRP) that prioritizes data packets transmission related to detecting COVID-19 over other data packets for timely and emergency measures. Simulation results show that the proposed PPRP routing protocol is more efficient in terms of end-to-end data transmission delay and network energy consumption than existing LEACH and CPWS protocols.

This paper introduces a real-time context-aware learning system that runs on mobile devices, collects data from the sensors, learns about the user-defined context, makes predictions in real-time, and manage IoT devices accordingly. However, the computational power of the mobile devices makes it challenging to run machine-learning algorithms with acceptable accuracy. Hence, existing works implement machine-learning algorithms on the server and transmit the results to the mobile devices. Although the context-aware predictions made by the server are more accurate than their mobile counterpart is it heavily depends on the network connection for the delivery of the results to the devices, which negatively affects real-time context learning. Therefore, in this paper, we propose a context-learning algorithm for mobile devices, which is less demanding on the computational resources and maintains the accuracy of the prediction by updating itself from the learning parameters obtained from the server periodically. Experimental results show that the proposed lightweight context-learning algorithm can achieve mean accuracy up to 97.51% while mean execution time requires only 11ms.

The Internet of Things (IoT) generates massive amounts of data everywhere through sensors of every kind which are disseminated in a variety of objects. This data contains incredibly valuable information useful for multiple applications. Knowing the context in which it was generated is extremely important and constitutes one of the first steps in extracting the knowledge it contains. Thereby, Context-Aware Learning (CAL) has become an important area of research as machine learning (ML) is a fast and ever-evolving technology. Wearable devices, ranging from accelerometers (ACC), frequently used, to magnetic field sensors, are used to monitor and recognize human activities (HA). Beyond ML Algorithms (MLA), accurate Human Activities Recognition (HAR) or context identification, depends not only on the kinds of sensors used but also on their location. In this paper, we study the impact of three types of sensors: ACC, gyroscope (GYR), and magnetometer (MAG); and their locations on the performance of MLA for HAR. Our results show that magnetic field sensors, less frequently used in the literature, placed at a specific location, provide the best performance in terms of HAR. Using a publicly available dataset, PAMAP2, we implement and evaluate the performance of HAR using five MLA: Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QLA), K-Nearest Neighbors (KNN), Decision Tree (DT), and Random Forest (RF). Our results show that the success rate of these algorithms is 98.3%, 90.4%, 97.6%, 99.9%, and 100% respectively, which exceeds the results obtained in a previous work based on the same dataset.

Wearable sensors are frequently used for monitoring physical activities and medical conditions. A variety of sensors are used, such as the accelerometer (ACC), gyroscope (GYR), and magnetometer (MAG), which are often embedded in Inertial Measurement Units (IMU). Data collected from these sensors can be used to identify context or physical activities through context-aware learning methods that apply a variety of learning algorithms. Implementing a real-time system (RTS) that serves a specific application like fall detection or heart condition, for example, is challenging as response time must be within a certain interval. This response time depends on the speed of data collection and transmission as well as the prediction time. Even though a lot of research was done in this area, to the best of our knowledge there is no comparative study based on the criteria we are using here. In this paper, we propose an Edge-based RTS for health-related applications and conduct a comparative study of several Machine Learning Algorithms (MLA) according to four criteria: source of data, sampling period, prediction time (PT), and success rates (SR). Even though MLA demonstrate different behavior toward these criteria, our simulation results showed that it is possible to implement a RTS that can identify or predict accurately physical activities within acceptable time constraints which are application dependant. Simulations were performed on wearable sensors’ data that we collected from 24 participants practicing five different physical activities. Our simulation results showed that the Decision Tree algorithm, with SR of 97.93% and PT of 0.17 seconds, outperformed all other algorithms.

IoT sensors are extensively used in a variety of medical applications such as patient monitoring. In recent years, multiple papers have been published on the measurement of blood alcohol level (BAL) using more specialized biosensors. Transdermal alcohol content (TAC), a practical and non-invasive tool for measuring BAL, was employed in several recent research works. As BAL affects the person's way of walking, accelerometers (ACC) combined with TAC data fed to machine learning algorithms (MLA) were applied to predict the state, drunk or sober, of a person. However, the accuracy of prediction was not high enough to be considered reliable. In this paper, using an archived “BAR CRAWL” dataset, we implemented five MLA: Linear Discriminant Analysis (LDA), Decision Tree (DT), Random Forest (RF), Extra Trees (ET), and Ada Boost (AB), with a variety of features to accurately predict the state of a person based on his alcohol levels. Furthermore, we defined various alcohol thresholds to predict the level of intoxication. Also, we were able to identify the intoxicated person. Our experimental results showed that we can achieve up to 28.9 % higher prediction accuracy with less processing time (PT) when compared to previous published works using the same dataset.

In User-to-User (U2U) communication in a Green Smart Cloud, a smartphone carried by a user can work as a transmitter. However, in some areas having only a handful of people, such as rural areas, smartphone-based transmitters can be scarce. Given that they should be placed at about every 100m2to be used for U2U communication, the placement of standalone transmitters at strategic points is significantly important. Moreover, estimating the minimum number of transmitters to cover the whole area to achieve network energy efficiency and quality of service (QoS) is crucial. This paper presents optimal redundant transmitter placement for U2U communication in a Green Smart Cloud to achieve QoS. Experimental results show that two redundancy, where at least two transmitters cover every point on the network, is more reliable than no redundancy and more cost effective than three redundancy, where at least three transmitters cover every point on the network.

In our society, the concept of sharing information is growing that lead to the emerging crowdsensing technology where a crowd of connected devices having sensors is used to sense and collect information. However, privacy is a true concern in crowdsensing technologies. Hence, this paper introduces two user's location privacy-preserving techniques based on anonymous reputation such as pseudonyms and different data collector/sender. Initially, we present very popular existing approach that uses pseudonyms to attach with the information to transmit data to the central server. Then, we introduce a simple but very effective approach DDCS, which uses different data collector and sender (i.e., one user collects information and swaps with other user that transmits collected information to the central server or platform). Finally, we introduce a hybrid approach that integrates pseudonyms with DDCS approach to achieve better privacy. Simulation results showed that hybrid approach outperforms pseudonym-based approach and DDCS in term of privacy preserving.

Contextual information can be learned at the mobile devices, such as smartphones, in real-time from the sensors to provide better services to the user. The sensor data collection process, where the data is collected by various internal sensors or autonomous external sensors, incurs greater power consumption, depending upon the type of sensor and data capturing rate in the mobile devices. On the other hand, the learning process itself drains the battery and at the same time affects the accuracy of the learning due to the limited computational power of the mobile devices. These problems can be addressed by shifting the learning process to the cloud, which is however achieved at the cost of reducing the accuracy of the real-time solutions and incurs heavy bandwidth usage depending upon the context of the user. Therefore, we propose a cloud-based real-time context-learning system where the user of the system will get the predetermined service in real-time according to the userdetermined context, which is learned from the related sensors while conserving a maximum amount of power compared to the standalone system or the cloud-based system. We have produced experimental results using a smartphone that illustrates that our system conserves 96.36% of power compared to the mobilelearning system while at the same time, the network data usage is 80% lower when compared to the cloud-based system. We have also showed that the proposed system works 76.17% and 94.81% faster compared to the mobile-learning and cloud-based system respectively.

Data sampling rate is an important performance metric in sensor-based environmental monitoring and real-time applications. Optimal and reliable monitoring in sensor-based applications requires frequent data sampling. However, when data is transferred through wireless channels, as is the case in Wireless Sensor Networks (WSNs), increasing this rate may limit battery lifespan. This is because Radio Data Transmissions (RDTs) are the most important source of energy consumption in WSN nodes. In this paper, we compare the performance of several RDT reduction algorithms such as Redundant Data, K-Means, Autoregressive data reduction (AR), Autoregressive Integrated Data Gathering (ARIMA) and Adaptive Distributed Data Gathering (ADiDaG), varying the data sampling rate, and determine the best sampling rate that minimizes a cost function, another performance metric. We perform the simulations using the readings from 20,000 real smart meters from the City of Moncton's Water and Utility Department. Simulation results demonstrate that the ARIMA approach achieves the best sampling rate that minimizes the cost function.

Road infrastructure is the life line of the transportation industry and it should be monitored at regular intervals to ensure that it provides a smooth riding experience, safety to the passenger and causes less damage to the vehicles. Road conditions are affected by several factors such as weather conditions, accidents that have occurred, and regular wear and tear, hence it is difficult to monitor them in real-time. Previous research works on road monitoring systems can be broadly classified into three groups; the first group uses sensor data to detect road conditions based on a given threshold, the second employs a machine learning algorithm to acquire sensor data from the vehicle, while the third uses machine learning at the server then transmits the result back to the vehicle. The learning algorithms of groups two and three provide better results than group one. Group three yields the more accurate results, but at the cost of time. Therefore, in this paper, we propose a novel system that monitors road conditions in realtime by learning from the data obtained from built-in sensors of a smartphone that is mounted inside the vehicle. We have designed a lightweight learning algorithm that improves accuracy by interacting with the server and monitoring road conditions in real-time. The algorithm is based on k-means clustering and our experimental results show that it can classify road conditions based on the accelerometer data with 88.67% accuracy.

Preserving connectivity in Mobile Wireless Sensor Networks (MWSN) having mobile nodes with heterogeneous mobility patterns is a great challenge and thus, is significantly important to ensure the success of mobility centric critical applications, like military, battlefield surveillance. This paper introduces a Dynamic, Heterogeneous Mobility-centric Clustering (DHMC) algorithm in which spatial and temporal densities are used. This protocol integrates the temporal aspect of a Minimum Period of Connectivity (MPC). Border nodes are used to reestablish connectivity among disconnected sub-clusters by increasing the communication range of the best suitable border nodes. Simulation was performed for three different mobility scenarios and showed that the proposed DHMC algorithm outperforms two well-known clustering algorithms LEACH and HEED in term of throughput and packets delivery ratio.

Intersections play an important role in traffic management. Traditional inte rsecti on management systems do not offer an optimal solution. Vehicular Ad-Hoc Networks (VANETs) offer communications among vehicles which allow the implementation of intelligent intersection management. Num erous techniques have been proposed in the li terature to regulate traffic at intersection intelligently. However, these research works do not provide a general solution based on user-defined priorities and vehicles density. In this paper, we propose a nove 1 priority based algorithm to man age intersection traffic us ing vehicle-to-vehicle communication and coalitions are formed among vehicles based on their priority. Our algorithm, let the vehicl es with higher priority pass the intersection before the lower priority vehicles. Simulation results showed that our algorithm improved throughput by 21.6% and travel time by 40.24% when compared with the traditional-based algorithm.

Recent studies have shown that radio data transmission is the most power consuming operation in power-constrained wireless sensor networks (WSNs). The power needed to transmit a single bit can be equal to that required for processing more than one thousand bits. Hence, increasing data processing to reduce data radio transmissions can considerably decrease power consumption. Several approaches have been designed to reduce data transmissions. This paper presents a novel approach, which combines partitioning clustering with the ARIMA algorithm. A performance comparison of the proposed approach with other data transmission algorithms, such as prediction using simple autoregressive and Adaptive Distributed Data Gathering (ADiDaG), is presented. Data from a real WSN, composed of more than 20,000 water meters in the City of Moncton, is used as a case study. Experimental results showed that the designed approach outperforms compared methods in terms of power reduction.

Technological advances have made Wireless Sensor Networks (WSN) more reliable, low cost, and widely used in a variety of applications. As WSN nodes are low power battery operated with limited lifespan, optimizing their resources (sensing, channel use, computing) is an essential task for the success of applications implemented over these networks. Even though this task could be challenging for large WSN, its impact in terms of network survivability and economic benefits can be significant. In this paper, we will focus on optimizing radio data transmission and evaluate its impact on power saving. We will analyze the data collected from a large WSN, more than 20,000 nodes, used for water meter readings in the City of Moncton. Two data driven approaches, reduction and prediction-based, are evaluated and compared. Then we propose a system for data collection and transmission that improves power consumption, in comparison with the existing one, without affecting the effectiveness of the application in terms of water consumption monitoring and detection of leaks. Experimental results show substantial power savings and significant increases in battery lifespan.

With the emergence of big data, designing an efficient distributed algorithm is significantly important. While most existing distributed algorithms consider distributed processing only for commodity computers, this paper introduces a Computationally Efficient, Dynamic distributed Algorithm (CEDA) for big data processing on a framework that comprises data processing both at the data collection end and data processing at the central server end. The proposed CEDA algorithm works both in low powered nodes and high speed commodity computers. Additionally, it performs sequential and parallel processing based on the amount of data received at the central server. Simulation results demonstrate that the CEDA algorithm achieves processing efficiency in terms of data processing time as compared to traditional distributed algorithms, which do not consider processing data at sensors.

Ensuring connectivity among mobile nodes is significantly important to the success of many critical applications, like military, where the loss of connectivity can lead to the failure of the mission and also proved to be challenging in particular when mobiles nodes have heterogeneous mobility pattern i.e., they are moving in several directions at different speeds. In this paper, we propose a new Heterogonous Mobility Pattern Clustering (HMPC) algorithm in which spatial and temporal densities are used. In our approach, we implement Density Based Spatial Clustering Application Noise (DBSCAN) in which the temporal aspect of a Minimum Period of Connectivity (MPC) is taken into consideration. Border nodes are used to reestablish connectivity among disconnected sub-clusters by increasing the communication range of the best suitable border nodes. Using three different mobility scenarios, simulation results show that the proposed HMPC outperforms two other typical clustering algorithms LEACH and HEED in term of connectivity as it provides better packets delivery ratio and smaller packets loss ratio.

Abstract Dynamic spectrum access policy allows a secondary user (SU) to access a primary user (PU) channel when it is idle. However, the idle state may result from the PU's silent activities, which can be wrongly perceived as an opportunity for the SU to access the channel and lead to “hidden collision” when it effectively tries to access the channel under this condition. At best of our knowledge, this issue has yet to be addressed in the literature. In this paper, we will first define a three‐state model that describes hidden collision conditions, then propose a probabilistic model in which a transient state is added to force the SU to wait a certain period of time before accessing the channel, thereby translating into better protection for the PU. Based on this model and using Carrier Sense Multiple Access/Collision Avoidance protocol, we will evaluate the PU's and SU's throughput and delay with and without transient state as well as the gain in using our approach. Our computation results show a substantial improvement of the PU's throughput from 19.6 to 61.1%.

Dynamic Spectrum Access (DSA) allows Secondary Users (SUs) to access a shared medium when it is in idle state, however, several challenges may be encountered. In this paper, we will focus on the Hidden Collision problem (HC) which occurs when Primary Users (PUs) are using Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) procedure, such as 802.11 protocol, and when SUs are trying to access the medium during the Backoff Window (BW) as they wrongly perceived the medium as available. As a consequence, PUs' performance may be negatively affected. In previous work, we proposed a new model that adresses this HC problem and allows better control of its effect. In this paper, we extend our work and study the effect of HC on PUs' performance in terms of throughput and delays under controlled and uncontrolled HC. Our results show that both throughput and delays decrease with the increase of SU transmission activity. The number of PUs has little effect on their throughput and a more significant impact on their traffic delays.

Vehicular Ad-hoc Network, VANET, offers new opportunities for better road traffic management. Using vehicle to vehicle (V2V) and vehicle to infrastructure (V2I) communications, several research works were published on road traffic improvement in general and at road intersection in particular. One of the challenging problems is to minimize vehicles' waiting time at the intersection as well optimize traffic flow by adjusting the traffic signals timing. In this area, several methods have been proposed in the literature which can be classified as adaptive, heuristic, pre-timed and others. In these methods traffic density and statistical data are generally used to determine the traffic signals timing. However, these methods either don't offer an optimal solution or are not cost effective in terms of computational time. In this paper, we propose the use of Simultaneous Perturbation Stochastic Algorithm (SPSA) to compute in real time the best traffic signals timing that minimizes the vehicles' travel time at the road intersection. A specific intersection in the city of Moncton with real traffic data was studied. Simulation results show a substantial improvement in term of vehicles' travel time in comparison with the pre-timed setting currently used by the city of Moncton. We also propose a heuristic algorithm which performs well too.

Wireless sensor networks (WSNs) have received significant research focus due to their widespread applicability in military, health care, agriculture, environment monitoring applications. These applications require WSNs to be self-organized and mobile. Clustering techniques are particularly important in the success of WSN implementations especially when sensor nodes are moving at different speeds. Many existing clustering algorithms assume that sensor nodes move at the same speed which is not true in many practical cases. Thus, in this paper, we introduce a velocity-based clustering algorithm and implement a relay placement technique in order to maintain seamless network connectivity. Simulation results show that the packet loss rate of the proposed algorithm is much lower than the existing LEACH and HEED clustering protocols.

In cognitive radio networks with primary users implementing CSMA/CA, such as 802.11 technologies, when all the primary users are in backoff states, the secondary users may detect primary users as inactive and try to access the channel, which will prolong primary users backoff procedure. This phenomenon is called hidden collision. In this paper, we first demonstrate how primary users' performance is degraded by hidden collision. Then, to protect primary users from excessive hidden collision, we propose a method to prevent secondary users from accessing the time slots during which all active primary users are in backoff states. Numerical examples are provided to show the effectiveness of the proposed solution.

This paper introduces a Mobility-centric Energy efficient and Fault tolerant Clustering protocol (MEFC) for Wireless Sensor Network that minimizes the number of cluster heads and active nodes to provide network coverage. Most existing routing protocols of WSN do not support mobile sensor nodes and are not fault tolerant. Considering the real world mobility-centric sensor applications the MEFC protocol allows the base station to control the mobility of sensor nodes so that they move to a new location to provide network coverage to be used in coverage-sensitive applications, e.g., battlefield surveillance. It also allows nodes to move inside or outside of a cluster and join a new cluster in data-centric applications, e.g., health monitoring for elderly people. The base station and cluster heads transmit beacon messages to discover the failure of cluster heads and member nodes of a cluster, respectively. Experimental results demonstrate that the proposed MEFC protocol reduces network energy consumptions and increased network lifetime as compared to the existing FT-EEC, DSC and LEACH protocols.

Rapid urbanization has put increasingly pressure on traffic management in urban areas. Conventional traffic signal with fixed or pre-defined variable cycles setting can slightly alleviate the increasing traffic problem, but cannot deal with continuously growing vehicular traffic in rapidly growing urban areas. VANET technology offers a promising solution for better vehicular traffic management in urban area to reduce traffic jam and improve transportation safety. Adaptive Traffic Light Control (ATLC) using VANET has attracted considerable attention from academic community. Unfortunately, most of these existing works used simulated traffic flow and hypothetical intersection architectures which may not reflect the reality of urban area. In this paper, we present a case study based on a specific intersection in the city of Moncton with real traffic data, and propose a new adaptive traffic light control algorithm. Our results show a substantial improvement of traffic throughput and average waiting time in comparison with fixed optimal cycle's time currently used by the city of Moncton and with existing adaptive solutions.

In this paper, we apply VANET capabilities to develop a new adaptive traffic light control algorithm that takes into consideration in real time traffic density and vehicle relative position to the intersection. Our preliminary results show a substantial improvement of traffic throughput and average waiting time in comparison with fixed optimal cycle's time and with existing adaptive solutions.

Abstract In cellular communication systems, in order for a network to keep track of inactive mobile stations (MSs), each inactive MS has to update its location from time to time, called location registration. To lighten the task of tracking inactive MSs, the network divides its cells into groups, called location areas (LAs) and tracks inactive MSs at the LA level: an inactive MS sends a registration message to the network to update its location only when it travels to a new LA. Obviously, the performance of a location area design depends on network traffic and the mobility of MSs. In the paper, we propose a general Markov mobility model for MSs in cellular networks, provide a procedure to automatically estimate the system parameters according to the network traffic, derive the performance of LA design, and provide a clustering algorithm to optimize LA designs. A numerical example is provided to show the effectiveness of the proposed procedures. Copyright © 2009 John Wiley & Sons, Ltd.

In order to save power to prolong battery life of subscriber stations (SSs) in IEEE 802.16e networks, the standard defines a sleep mode for SS. When there is no traffic for an SS to transmit or to receive, the SS switches to sleep mode periodically. The sleep interval is doubled each time until a maximum sleep interval threshold Tmaxis reached. Obviously, the performance of this power saving mechanism depends on the idle period distribution, which is user-specific. In network traffic modeling, it is commonly accepted that frame interarrival times have heavy-tailed distributions. Since heavy-tailed distributions make analysis and design challenging, in this paper we propose to use mixtures of exponentials to approximate heavy-tailed idle times. With a mixture of exponentials approximating the idle times, performance can be explicitly derived and optimized. An online EM algorithm is proposed to fit the mixture of exponential distributions to the idle times. Numerical examples show the effectiveness of the proposed procedures.

Smart antenna is a promising technology to improve the capacity of wireless networks. Implementing smart antenna in WLAN urgently demands new multiple access protocols. In this paper we propose a multiple access protocol for WLAN to effectively support smart antenna. The proposed protocol makes use of the hybrid superframe structure of polling and contention, which is illumed from IEEE 802.11, and distinguishes the stations residing within the AP's broadcast range from those out of the broadcast range. For the stations residing within the AP's broadcast range, the AP uses polling to transmit in broadcasting mode; for those residing out of the AP's broadcast range but in its directional range, the AP uses contention-based method to transmit in beamforming mode. We have evaluated and optimized the protocol performance against its parameters by simulation. To show the efficiency of the proposed protocol, we also compare it with other protocols in average packet delay and throughput. The simulation results show that the proposed protocol performs well; especially, under heavy traffic load, it achieves the smallest average delay and the highest throughput.

The use of smart antenna to extend the coverage range and capacity of wireless local area networks (WLAN) dictates the employment of novel multiple access protocols, with which the access point (AP) can provide access to remote stations. In this paper, we propose an adaptive multiple access protocol for WLAN with smart antenna. To efficiently support the differentiation of the stations residing within the broadcasting coverage range of the AP from those residing out of the range, the proposed protocol uses the hybrid superframe structure of contention-free period and contention period. Moreover, our protocol can adaptively adjust the antenna mode between beamforming and broadcasting according to the beam width and the number of active stations in the service area. Using simulation we compared the adaptive protocol with other non-adaptive ones. The results show that constructing our adaptive protocol by switching among non-adaptive protocols turns out to be the best strategy which allow us to take advantage of the best protocol providing low delay and high throughput.

tiplexer (see 157, 1211. This paper applies integer-vatuecl time series to characterize the mffic autocorrelation and Atrroronaiorinn and mapinal ulisrrihtrrinn are hlto im- rnaqjna1 distribution *fthe packet a h~t in ra.w of ve~y /OM' EO.P.T rare, ui~for~~~elat~m i.~ more rriticul.

The hyper-Erlang model can approximate any non-negative distribution as closely as desired. It can not only characterize field data well, but also facilitate analytically tractable results for performance evaluation. As a result, this model is widely used in telecommunication network modeling. This paper proposes an online algorithm for Gamma mixture distributions, which contain hyper-Erlang models as a special case, and applies the algorithm to Internet traffic modeling. Simulation and experimental results are also provided.

In this paper, we propose a Markov movement model for mobile terminals in wireless personal communication service networks and study the location registration problem. We formulate the location registration as a Markov decision process and prove that the optimal strategies have threshold structures. To avoid solving the Bellman dynamical programming equations, we propose a single sample path-based algorithm to tune thresholds of the strategies. Since the proposed algorithm uses only one sample path of the system, it can be implemented online