Accurate short-term load forecasting (STLF) is essential for Internet of Things (IoT)-enabled smart buildings. This paper identifies theglobal-error couplingproblem in the conventional multi-input multi-output (MIMO) forecasting strategy, which limits near-term accuracy. To address this issue, we propose Hierarchical Step Fusion (HSF) to decouple multi-horizon optimization and Adaptive Model Weighting (AMW) to combine horizon-specific models. Experiments on real-world building data show that the proposed framework achieves 36% lower root mean square error (RMSE) than the best baseline for the first-step prediction (k= 1) under an 8-step forecast horizon (H= 8) and up to 20% lower RMSE overall, enabling more accurate and sustainable IoT-based energy management.

Parkinson’s disease (PD) is a prevalent neurodegenerative disorder that affects motor function and impacts millions worldwide. Early diagnosis and precise staging of PD are critical for effective management and timely intervention. Traditional gait analysis focuses on lower limb movement to differentiate between moderate and advanced PD stages; however, integrating upper and lower limb dynamics offers a more comprehensive approach to distinguishing early-stage PD from other. This study introduces a novel spatio-temporal model that integrates CNN, Transformer, and BiLSTM layers to classify PD based on the Hoehn and Yahr (H&Y) score. The architecture captures both spatial and temporal dependencies in gait patterns, with a CNN serving as the primary spatial feature extractor, utilizing a temporal-focused kernel, and augmented by a Transformer to enhance temporal feature learning. Additionally, a BiLSTM module synthesizes hierarchical spatio-temporal features through multi-path integration of CNN and Transformer outputs to model complex gait dynamics. For data augmentation, sliding windows of 500, 800, and 1200 samples were employed, and the model’s performance was rigorously assessed through 3-, 5- and 10-fold cross-subject validation. Metrics included accuracy, F1-score, recall, precision, and a voting-based confidence score, while t-SNE visualization provided insights into spatio-temporal feature differentiation. The model demonstrated high performance, achieving optimal accuracy of 0.90 and an average confidence score of 0.97 with a 1200-sample window in 5-fold validation. These findings underscore the potential of spatio-temporal deep learning architectures to advance multi-class PD classification and highlight their effectiveness in detecting early-stage PD from gait pattern signals.

As network technologies evolve, accurate internet traffic prediction has become vital for efficient resource management. This paper introduces a novel ensemble - learning framework that integrates Temporal Convolutional Networks (TCNs) with a No Negative Constraint Theory (NNCT)-based weighting strategy. Unlike traditional approaches that rely on simple averaging or conventional weight constraints, our method uses the limited-memory Broyden-Fletcher-Goldfarb-Shanno algorithm with box constraints (L-BFGS-B) to optimize weight coefficients precisely within predefined bounds, thereby avoiding the instability often associated with metaheuristic algorithms. Experimental results demonstrate significant improvements in both prediction accuracy and robustness, particularly during periods of high demand. Our approach outperforms current leading forecasting techniques, validating its effectiveness and its potential to enhance network management.

In response to the critical need for energy conservation and carbon emission reduction globally, this paper introduces an innovative framework for hourly energy forecasting in smart buildings. Utilizing data collected from sensors, this framework forecasts future energy consumption, enabling efficient monitoring and control of electrical appliance usage through advancements in the Internet of Things (IoT). This reduces energy waste significantly. Diverging from traditional statistical models, which often falter with nonlinear power load sequences, our approach employs a Temporal Convolutional Network (TCN). This robust framework is designed to improve accuracy, robustness, and generalization in forecasting energy usage in buildings. It incorporates an adaptive method to enhance the practical applicability and dynamic updating of our proposed solutions in real-world scenarios. A comparative analysis with existing energy forecasting methods demonstrates the superior performance and adaptability of our framework, showing a reduction in error of more than 20%. This study aims to establish a new benchmark in energy forecasting for IoT-enabled smart buildings, offering a scalable and precise solution for energy consumption management aligned with global sustainability goals.

With the advancement of artificial intelligence (AI), indoor positioning systems have become increasingly important for various applications, leading to the development of diverse indoor positioning methods. Among various approaches, image-based indoor positioning methods have demonstrated relatively good positioning accuracy. However, maintaining user privacy continues to be a significant challenge. To tackle this issue, we propose a Federated Learning (FL) approach and develop an FL-based indoor image recognition positioning system. We conducted experiments to validate our method using two real-world datasets and compare the results to a Non-FL approach. Furthermore, we evaluated the accuracy of our method across various communication rounds, the number of client devices, and the amount of data per client device. Our experimental results indicate that the proposed method effectively preserves client privacy while achieving accuracy similar to the Non-FL approach. By employing FedAvg and FedOpt algorithms with the MobileNet model, our system ultimately reaches 94% accuracy.

In recent years, the rapid advancement of deep learning has enabled researchers to locate individuals indoors using image recognition techniques. As a result, we propose an indoor positioning method that combines image and machine learning techniques. This study utilizes the robust OpenPose pose estimation model to accurately identify human joints and extract the position of the individual's foot from an image. Then, we apply the 2D direct linear transformation approach to determine the coordinates of the individual's foot position in the indoor environment. To reliably determine the position of the foot in complex indoor spaces where potential obstructions often occur, we propose a novel foot position extraction algorithm. Finally, we employ classical machine learning models, such as linear regression, to reduce distance errors caused by lens distortion and other factors. Our proposed method achieves highly accurate indoor positioning with an average distance error of only approximately 0.4 meters, just by using a single camera.

Deep learning models that exceed human capability are also pruned to error when facing adversarial attacks. Recent works on adversarial reprogramming have unleashed the capability of repurposing a machine-learning model for another task without changing the model parameters. In this paper, we prove the possibility of repurposing a single machine-learning model to solve multiple tasks by adding adversarial-reprogramming functions. In addition, adversarial reprogramming can also be a solution in a memory-limited device that saves more space by storing only one model instead of multiple task-specific models in a single device. In our experiment, adversarial reprogramming can achieve 95% accuracy prediction from a model trained on a different domain. Even in the cross-model type, adversarial reprogramming can perform 83% accuracy on sentiment analysis of a sentence with a model that receives an image as its input type. We also did the experiment on a low-resource language, Bahasa Indonesia, and achieve 75% accuracy without changing the model parameter

Classifying epitopes is essential since they can be applied in various fields, including therapeutics, diagnostics and peptide-based vaccines. To determine the epitope or peptide against an antibody, epitope mapping with peptides is the most extensively used method. However, this method is more time-consuming and inefficient than using present methods. The ability to retrieve data on protein sequences through laboratory procedures has led to the development of computational models that predict epitope binding based on machine learning and deep learning (DL). It has also evolved to become a crucial part of developing effective cancer immunotherapies. This paper proposes an architecture to generalize this case since various research strives to solve a low-performance classification problem. A proposed DL model is the fusion architecture, which combines two architectures: Transformer architecture and convolutional neural network (CNN), called MITNet and MITNet-Fusion. Combining these two architectures enriches feature space to correlate epitope labels with the binary classification method. The selected epitope-T-cell receptor (TCR) interactions are GILG, GLCT and NLVP, acquired from three databases: IEDB, VDJdb and McPAS-TCR. The previous input data was extracted using amino acid composition, dipeptide composition, spectrum descriptor and the combination of all those features called AADIP composition to encode the input data to DL architecture. For ensuring consistency, fivefold cross-validations were performed using the area under curve metric. Results showed that GILG, GLCT and NLVP received scores of 0.85, 0.87 and 0.86, respectively. Those results were compared to prior architecture and outperformed other similar deep learning models.

The demand for safe internet access will increase in the future, as a variety of malicious attacks are targeting fragile devices in the network. For this reason, a network intrusion detection system (NIDS) is utilized to detect and notify attacks. Various algorithms employing pattern analysis techniques have been successfully implemented in NIDS, such as shallow machine learning and deep learning models. However, the problem of an imbalanced dataset is often negligible in intrusion detection. Often, the results achieved by machine learning tend to majority classes and ignore minority classes. Therefore, we propose the focal loss network intrusion detection system (FL-NIDS), an improved classification model, to address the issue of an imbalanced dataset. This paper focuses on finding the most appropriate hyperparameter values for focal loss in intrusion detection systems. To show the effectiveness of the proposed FL-NIDS, we utilize the UNSW-NB15 dataset, which suffers the imbalance classes. Our experiment shows that the FL-NIDS using the best combination parameters of focal loss obtain higher detection accuracy in the majority classes as well as in the minority classes, compared to the general model such as Deep Neural Network (DNN) and Convolutional Neural Network (CNN).

Multiuser massive multiple-input multiple-output (MU-MIMO) systems can be used to meet high throughput requirements of 5G and beyond networks. In an uplink MU-MIMO system, a base station is serving a large number of users, leading to a strong multi-user interference (MUI). Designing a high performance detector in the presence of a strong MUI is a challenging problem. This work proposes a novel detector based on the concepts of expectation propagation (EP) and graph neural network, referred to as the GEPNet detector, addressing the limitation of the independent Gaussian approximation in EP. The simulation results show that the proposed GEPNet detector significantly outperforms the state-of-the-art MU-MIMO detectors in strong MUI scenarios with equal number of transmit and receive antennas.

A deep neural network (DNN) has the ability to rival humans in performing wide variety of tasks. The exceptional ability of DNN comes with high computational, memory, and power cost. In order to enable DNN on edge devices and any realtime environments like self-driving car, many techniques have been developed. Most of the techniques use a single network and sacrifice huge amount of accuracy to reduce the cost. In our paper, we propose a hierarchical network to enable computation on edge devices with the help of a cloud to reduce the accuracy loss. The networks have to be able to detect uncertainty in their prediction which is not usually calibrated. This paper uses modern portfolio theory to calibrate uncertainty on the model prediction. The idea is to optimize the return of a horse race problem based on the doubling rate of gambling where a gambler needs to find the proportion between making a prediction and abstain when the network is not confident. For the high abstain score, the data will be forwarded to a deeper model for better prediction. We demonstrated that the proposed framework outperforms knowledge distillation and pruning algorithms. The superiority of our result enables a high-quality result on intelligence IoT applications.

Based on recent developments in speed estimation and deep learning technology, we prefer to use only one camera for speed estimation to reduce hardware costs, and calculate the speed immediately on the screen instead of calculating the average speed over a distance, making it easier to install the device. We use a machine learning method to detect vehicle objects, so as not to be limited by the direction of the vehicle. In addition, the road traffic marking rules can be used to obtain the size of all the road markings, which can be converted to a calibration matrix by road marking-based perspective transformation. Therefore, there is no need to measure the size of the road or get the scale of the image beforehand to calibrate and then complete the speed estimation. An average error is 3.2 km/hr, average relative error is 9.94%. In the future, we plan to improve the recognition accuracy and make the system lighter and cheaper by building it in an embedded system.

Global Navigation Satellite Systems (GNSS) have been widely used in outdoor navigation. However, the problem of signal attenuation and shadowing effect is serious in an indoor environment. Hence, an accurate smartphone based multi-floors Indoor Positioning System (IPS) that can guide users through indoor environments, where similar corridors and rooms across multiple floors are commonplace, is highly desired. In this paper, we have implemented an indoor positioning system using an enhanced Pedestrian Dead Reckoning (PDR) scheme without any external sensors on commercially available smartphones. And it is evaluated in a real practical indoor environment that simulates normal user traveling routes through multiple floors and corridors, conducted by several users with different strides and walking patterns in order to evaluate its performance across different individuals. The evaluation result showed our system can achieve a mean accuracy 2. 59m in a multi-level building including an 80mx16m area on each floor without any floor estimation error.

This work aims to implement a real-time scene identification system using an image captioning model on an embedded system. The image captioning model can translate the image captured by a webcam installed on the embedded system into a human-readable sentence immediately. Users can get the information quickly by reading only the sentences. There are two stages in the image captioning model. First, a deep neural network extracts features from images captured from the webcam. Second, a long-short term memory generates the corresponding sentence. Due to the portability of the embedded system, our scene identification system can be placed anywhere at home or in the company. We evaluate the execution time in different aspects on several embedded systems and demonstrate the generated sentences from the captured images by our scene identification system.

The Global Positioning System (GPS) is widely applied for outdoor positioning. However, GPS signals may degrade in an indoor environment. A nursing cart with a stable and accurate indoor positioning system inside is highly demanded in the hospital. The cart can be used for tracking the employees, or positioning movable medical devices. In this paper, we propose a Pedestrian Dead Reckoning (PDR) -based indoor positioning system with iBeacon calibrations. The proposed system can be implemented in resource-limited embedded device. To evaluate the property of the proposed system, real experiments under indoor environments have been conducted. The experimental results showed the effectiveness of the proposed approach. We also compared the impact of the number of iBeacons on positioning accuracy.

The rapid development of Indoor Positioning System has attracted researcher to develop a robust scheme to predict the location based on Received Signal Strength Indicator (RSSI) signal. A lot of research topics presented in many journals and conferences by many researchers concern indoor positioning system as a main topic [1], [2]. Currently, the study related to find the robust algorithm for indoor positioning system becomes a high demand topic in several conferences. Our work intents to evaluate the effectiveness of Recurrent Neural Network (RNN) as a deep learning technique to be implemented in this field. In addition, LSTM as a variant of RNN scheme is also implemented. The purpose of this implementation is to explore both LSTM and original RNN to be utilized for localization in indoor positioning scheme, especially for Wifi Fingerprinting Dataset. From all evaluations, our proposed approach could get 99.7% accuracy for predicting which floor the sensor belongs to. In addition, the distance errors of our scheme are around 2.5-2.7 meters.

Bike sharing systems have been widely applied to many cities and brought convenience to local citizens for short-ranged transportation. The bike shortage problem due to uneven bikes distribution is one of the biggest challenges in bike sharing systems. In this paper, we focus on station level prediction for each bike station. The proposed architecture is based on Recurrent Neural Network (RNN) and we use only one model to predict both rental and return demand for every station at once which is efficient for online balancing strategies. Without considering the global level bike distribution, the MAPE/RMSLE of the sum over the demand of each station may be too high for rebalancing strategies but the MAE/RMSE are satisficing at station level. Our evaluation shows that the proposed methods meet satisfied results at station level and global level in New York Citi Bike dataset.

Energy consumption is a major concern in Aerial Wireless Sensor Network(AWSN), especially when sensors are mounted on airborne platforms with limited power supply. Efficient energy aware clustering scheme can help them gathering and transmitting as much data as possible by extending their limited operation lifetime. In this study, we focus on analyzing how mobility models in 3-D space affect our previous work - Regional Energy Aware Clustering with Isolated Node (REAC-IN) scheme, since sensors moving in 3-D environment bring up the issue of isolated node problem even further. The simulation result shows the implementation of REAC-IN under the scenario of AWSN still performs well.

Nowadays people can easily use any smart device at hand to capture scenery and then upload it to the cloud storage. Cloud storages are widely used for storing the generated multimedia content. However, the risk of potential private data leakage may exist since cloud storages are normally in a public domain. To enhance the security and privacy of images on the cloud storage, we proposed an integrated scheme evolving invisible digital watermarking and masking which are based on the histogram shifting method. The histogram modification based scheme can achieve reversible data hiding, to ensure the integrity and the confidentiality of the image data. Additionally, we use the secret sharing scheme to keep the secret keys to further improve the security of data access. The evaluation results show that the proposed system can effectively prevent the malicious user from accessing the private images.

Information searches are the most common application within social networks. Normally, the social network is modeled as a network graph, consisting of nodes (In the rest of the paper, unless otherwise specified, we will use the terms “user” and “node” interchangeably.) representing users within the network and edges representing relationships between users. Choosing the appropriate nodes to form an auxiliary structure for supporting the effective query message spreading can reduce the troublesome repeated queries. To accomplish this, a hybrid search (HS) scheme is proposed. If the query message is received by a node belonging the auxiliary structure constructed by dynamic weighted distributed label clustering (DW-DLC), it would be flooded to all neighbors of the visited node; otherwise, it would be forwarded to one neighbor of the visited node. The DW-DLC based auxiliary structure can accelerate the process of obtaining required information within the network. The simulation results show that the HS+DW-DLC scheme can reduce the average searching delay time, even in a required-information-scarce social network. In addition, the proposed scheme can generate a relatively low amount of repeated messages to lower repeatedly asking social network users.

Energy consumption is an important issue in the design of Wireless Sensor Networks (WSNs) which typically are powered by limited energy. A suitable clustering algorithm for grouping sensor nodes can raise energy efficiency for WSNs. However, there exist some overheads for clustering, such as cluster-head selection, assignment, and cluster construction. In this paper, we propose a new regional energy aware clustering with isolated nodes for wireless sensor networks, called REAC-IN. In REAC-IN, the cluster-heads are selected by the calculated weight based on the residual energy of each sensor and the regional average energy of all sensors in each cluster. A distributed clustering algorithms may result in isolated nodes which are far away from cluster-heads. Such isolated nodes may need to communicate with the sink directly by consuming much energy. In order to reduce the energy consumption and prolong the network lifetime, we use regional average energy and the distance between sensors and the sink to determine whether the isolated nodes sleep or transmit data. Our simulation results validate that REAC-IN can outperform other clustering algorithms, such as LEACH, HEED and DEEC algorithms, and has better performance in terms of the total amount of transferred data, the total network energy consumption, and the network life-time.

Each sensor node in a wireless sensor network (WSN) mainly consumes energy to sense the environment and convey or relay the sensed data to a sink node. Once the resident energy in a sensor node is drained, this may cause incomplete sensing coverage, resulting in WSN failure. The lifetime of a network begins when the network starts working and ends when the first node becomes ineffective from energy exhaustion caused by any of the sensor nodes in the WSN. This paper proposes a scheme that clusters sensor nodes by node reading and by node location. By using the resident energy and coverage of each node, the scheme proposes a weakest node protection mechanism to balance the energy consumed by each node, to extend the network's lifetime. The evaluation results validate the proposed concept and show the proposed scheme can even obtain additional 12.5% network lifetime compared to other existing ones.

In a distributed environment, a fundamental concern is authentication of local and remote users in insecure communication networks. Absolutely, legitimate users are more powerful attackers, since they possess internal system information not available to an intruder. Therefore many remote user authentication schemes for distributed systems have been proposed. These schemes claimed that they could resist various attacks. However, they were found to have some weaknesses later. Lee et al . proposed a secure dynamic ID‐based remote user authentication scheme for the multi‐server environment using smart cards and claimed that their scheme could protect against masquerade attacks, server spoofing attack, registration server spoofing attack and insider attack. In this study, the authors show that Lee et al . 's scheme is still vulnerable to password guessing attack, server spoofing attack and masquerade attack. To propose a viable authentication scheme for distributed systems, we remedy the flaws of Lee et al . 's scheme and propose an efficient improvement over Lee et al . 's scheme. Furthermore, we compare the proposed scheme with related ones to prove that the computation cost, security and efficiency of the proposed scheme are well suitable for practical applications in a distributed system.

With the emergence of WiFi technology and network-based applications, the computing, communication and sensing capabilities of smartphones are increasing rapidly, and the smartphone has emerged as a particularly appealing platform for pervasive network applications. However, WiFi entails considerable energy consumption on these battery-powered devices. Finding ways to reduce power consumption on smartphones becomes a critical issue. In this paper, we propose an adaptive limit-rate selection algorithm based on anon-parametric signal strength prediction scheme and analyze its potential for energy savings. By periodically monitoring the received signal strength (RSS) in diverse network environments, the proposed scheme applies weighted scatter plot smoothing and kernel moving average algorithms to adaptively adjust file downloading and video streaming rates. Experimental results demonstrate that the proposed scheme can save 5.7% energy at least and 13.9% energy at most compared to non-adaptive and non-prediction schemes when the smartphone holders use the applications on the move.

An Internet of Things (IoT) system features integration information from heterogeneous sensor devices, allowing them to deliver a variety of sensed information through networks. An IoT broker in the system acts as an information exchange center, relaying periodic messages from heterogeneous sensor devices to IoT clients. As more devices participate in the IoT system, the service scale of entire system increases. To overcome the limitation of the number of direct links to an single working element, the whole IoT system should be divided into many subsystems, called IoT units, to form the system hierarchy. Furthermore, applying the service-oriented architecture (SOA) concept to realize the IoT service can facilitate to adapt the future-proof devices to the IoT system. Normally, a web-based message visualization can unify the client interface in a SOA-based system. However, a large volume of web-based messages with various sizes are not easy to be stably displayed at the client side. This paper proposes an IoT system skeleton and a shortest processing time (SPT) algorithm for scheduling web-based IoT messages. The implemented scheduling scheme supported by a priority queue model can effectively stablize the response messages from the scattered IoT sensors per each client request.

Anonymous channel tickets have been proposed as a way to provide user anonymity and to reduce the overhead of re-authentication for authentication in wireless environments. Chen et al. proposed a secure and efficient protocol, based on a protocol proposed by Yang et al., which is resistant to guessing attacks on networks from which users’ secret keys are easy to obtain. However, their scheme is time-consuming in the phases of ticket issuing and authentication. Furthermore, a malicious attacker can utilize the expired time, Texp, to launch a denial of authentication (DoA) attack, which is a type of denial of service attack. Because Texp is exposed to any user, it would be easy to launch a DoA attack that could make the scheme impractical. To resist against DoAs that the scheme of Chen et al. might suffer, we propose an improved scheme based on elliptic curve cryptography in this paper. Our scheme not only reduces time cost but also enhances security. The basis of the proposed scheme is the elliptic curve discrete logarithm problem. The operations of points of an elliptic curve are faster and use fewer bits to achieve the same level of security. Therefore, our scheme is more suitable for mobile devices, which have limited computing power and storage. Copyright © 2012 John Wiley & Sons, Ltd.

Information searching is the most common application in the social network. Choosing appropriate users as the clusterheads in a clustering based social network can facilitate to effectively spread query messages, resulting in efficient searching performance. To do so, we propose a hybrid searching (HS) scheme and use the dynamic weighted distributed label clustering (DW-DLC) structure to assist the searching process. The DW-DLC structure can accelerate the process of finding target information in the social network. The simulation results depict that the HS+DW-DLC scheme can reduce the average searching delay time by more than 90% even in a information-scarce social network which only 8% of total nodes can provide the target information. Meanwhile, HS+DW-DLC can generate a lowest average message overhead, which may burden the network, compared to other schemes.

Wireless sensor networks (WSNs) represent the next evolutionary development step in utilities, industrial, building, home, shipboard, and transportation systems automation. WSNs are easy to deploy and have wide range of applications. Therefore, in distributed and unattended locations, WSNs are deployed to allow a legitimated user to login to the network and access data. Consequently, the authentication between users and sensor nodes has become one of the important security issues. In 2009, M. L. Das proposed a two-factor authentication for WSNs. Based on one-way hash function and exclusive-OR operation, the scheme is well-suited for resource constrained environments. Later, Khan and Algahathbar pointed out the flaws and vulnerabilities of Das's scheme and proposed an alternative scheme. However, Vaidya et al. found that both Das's and Khan-Algahathbar's schemes are vulnerable to various attacks including stolen smart card attacks. Further, Vaidya et al. proposed an improved two-factor user authentication to overcome the security weakness of both schemes. In this paper, we show that Vaidya et al.'s scheme still exposes to a malicious insider attack that seriously threatens the security of WSNs. We hope that by identifying this vulnerability, similar schemes can avoid these weaknesses.

In this video demonstration, we present a remote interaction user interface prototype -- Pointer Wizard, to provide mixed reality experience with intuitive interaction afar. By identifying finger gestures and voice commands, it allows users to interact with appliances remotely, without the need of close proximity or even physical contact. It also brings information from the virtual world out into the real surrounding, thus creating the mixed reality experience. Due to the rapid improvement and development of smartphone and smart devices, we can now integrate the readily available hardware with micro-projection technology and quickly create prototypes that can be deployed rapidly. We designed image and speech recognition apps for smart devices in order to identify the figure gestures and target appliances, and through wireless transmission not only sending control signals to control relays for the appliances and receiving feedback information, but also gathering related information from the Internet,. The prompt control commands and related information are then projected out via portable light-weight projecting mechanism in real-time, hence users are seemingly able to operate and interact with the target from a distance. Through realizing this project, we hope to provide a more convenient way of life, and a more intuitive way of obtaining information or manipulating home appliances, like performing magic as wizards in a fantasy.

Power saving in mobile devices has become a hot topic nowadays. Enhancing user experience through intelligent techniques that help to extend battery lifetime is today's goal for many manufacturers and developers. Different approaches to improve power consumption have focused on improving the hardware used, the operating system, and/or the applications performance; however, this study's focusisto design an intelligent software framework. After a throughout evaluation of the power consumption in different modules of the phone, in order to identify which modules consume power the most,a reinforcement learning method is proposed to effectively deal with this issue by granting/denying accessto the user of executing battery-draining tasks.

Cloud computing features a flexible computing infrastructure for large-scale data processing. MapReduce is a typical model providing an logical framework for cloud computing and Hadoop, an open-source implementation of MapReduce, is a common platform to realize such kind of parallel computing model. Normally, a cloud computing service comprises many heterogeneous commodity machines. The original resource arrangement policy in Hadoop only focuses on the logical resources, such as free slot number, without considering the physical workload of comprehensive computing resources, such as the CPU utilization, network bandwidth, memory usage on each working node. This paper aims at dispatching the computation load to all processing nodes in the cloud computing environment by considering the physical workload on each node so as to prevent bias in arranging computation resources and hence improve the overall computing performance in a heterogeneous cloud environment.

Conventional IEEE802.15.4 wireless sensor network is suitable for fixed sensor nodes. However, there are some issues with mobile sensor nodes joining or leaving the sensor network since the mobile sensor node is easier to lose its connectivity compared to the fixed sensor nodes. In order to regain its connectivity, the mobile sensor nodes must spend more time and energy to perform the re-association procedure, resulting in an efficiency decay of the sensor network. To solve this problem, we refer to the previous research to design an adaptive weighted link quality indicator prediction based scheme in the media access layer. The proposed scheme can reduce the non-connectivity time of mobile sensor node by early performing the re-association procedure. Additionally, we design a simple procedure to lessen the Ping-Pong effect problem without additional computation overhead, thereby improving stability of the connection between the mobile sensor node and its corresponding coordinator.

Positioning is a basis for providing location information to mobile users. With the GPS signal strength hindered by the structure, indoor users cannot obtain their positions through their handheld devices. Indoor positioning has recently been conducted based on received signal strength (RSS) fingerprint from indoor wireless access devices. Integrating indoor position information with Smartphone can provide more precise in-building information. In this paper, we propose a novel indoor positioning scheme assisted by a RSS fingerprint and footprint architecture. Smartphone users can get their position based on RSSs from ambient Wi-Fi access points surrounding them. With the assistance of collecting RSSs from ambient Wi-Fi signals, confining RSSs by directions, and filtering burst noises, our propose scheme can overcome the severe signal fluctuation problem in the building. Meanwhile, the proposed RSS fingerprint and footprint matching mechanism can raise the accuracy of location estimation. The experiment results show that our proposed scheme can achieve a certain level of accuracy in the indoor environments and outperform other solutions.

This paper proposes a new algorithm, TBRA (Termites- Based Routing Algorithm), for use in three- dimensional mobile wireless sensor networks. The TBRA is a routing selection strategy based on the concept of the ant colony optimization algorithm. It is difficult to establish and maintain optimal routes in a 3D space because the RSSI, energy of sensor nodes, vary with time. This routing selection strategy can be used for various traffic conditions by measuring and counting the average energy of nodes, RSSI value. The main contribution of this study combines ACO algorithm and Durkin's propagation model to derive an efficient routing scheme for wireless sensor networks in a 3D space.

As the mobile networks are springing up, mobile devices become a must gadget in our daily life. People can easily access Internet application services anytime and anywhere via the hand-carried mobile devices. Most of modern mobile devices are equipped with a GPS module, which can help get the real-time location of the mobile device. In this paper, we propose a novel authentication scheme which exploits volatile passwords - One-Time Passwords (OTPs) based on the time and location information of the mobile device to transparently and securely authenticate users while accessing Internet services, such as online banking services and e-commerce transactions. Compared to a permanent password base scheme, an OTP based one can prevent users from being eavesdropped. In addition to a memoryless feature, the scheme restricts the validness of the OTP password not only in a certain time period but also in a tolerant geometric region to increase the security protection. However, if a legitimate user is not in the anticipated tolerant region, the user may fail to be authenticated. Hence, a Short Message Service (SMS) based mutual authentication mechanism is also proposed in the article to supplement the unexpected misjudgement. The proposed method with a volatile time/location-based password features more secure and more convenient for user authentication.

In a cooperative network environment, each independent node communicates with others by some agreement mechanism to make sure the majority of nodes can actuate a common corresponding action. Byzantine Agreement (BA) has been proved to effectively make all healthy nodes obey the same command issued from the commander node. The major challenge for BA is its considerable amount of exchange messages among nodes to reach the consensus, especially when BA is applied to a network environment which comprises lots of cooperative nodes. Based on above, we propose a Virtual Grouping Byzantine Agreement (VGBA) scheme to compromise between the percentage of nodes which can obtain the common command and the number of exchange messages. VGBA may suffer a lowered tolerance capability of faulty nodes and a lowered percentage of healthy nodes having the command but significantly reduce the message volume so that the improved scheme is applicable while facing a larger number of nodes. The theoretical inferences and evaluation results show that the proposed scheme provides an effective trade-off solution to the agreement problem among cooperative distributed nodes in terms of message volume and fault tolerance in an autonomous network.

Under the beyond third-generation (B3G) vision, cellular operators (like Vodafone, T-mobile) construct a complementary environment – public wireless local area network (PWLAN) to compensate for the limited data access service to become cellular/PWLAN operators. In addition, PWLAN service coverage can be improved by bringing together various types of PWLAN operators, including PWLAN operators, cellular/PWLAN operators and PWLAN roaming brokers under PWLAN roaming agreements. To build a better PWLAN access service, the group formed by heterogeneous PWLAN roaming partners is markedly growing. In such a cooperative service, the user authentication process plays a foremost role for users and involves many back-and-forth authentication message transmissions among the PWLAN partners. Some factors causing the authentication delay are hard to be further improved, such as the authentication network condition since the networks are owned by many PWLAN partners and scattered in the Internet. If the Internet protocol (IP) lookup time for the authentication server can be shortened, the efficiency for the IP-based authentication process can be made. In this article, an improved mechanism for reducing the IP lookup latency prior to conveying authentication messages is presented. With theoretical analyses and simulation results, the proposed scheme can improve the authentication process in a cooperative B3G network.

A traditional wireless sensor network is composed of several sensors and a sink. The sink analyzes data measured by the distributed sensors and takes appropriate action. A problem with this kind of architecture is that it may have a single-point of failure. Also, sensors are not connected directly to the sink and must send data by hopping through other sensors. This means that it would take more time for the sink to collect data. In a wireless sensor network, noise may distort the message during transmission. An intruder may also alter the message maliciously. So far, there has been little research done on the design of robust wireless sensor networks to overcome the single-point of failure problem and environmental interference. In this study, we propose a consensus problem algorithm based solution to enhance the accuracy of the detected result in an autonomous local sensor network without a centralized sink. Under our scheme, there is no need to send the detected value to the sink. The solution can therefore reduce the transmission and routing time, allowing appropriate action to be made directly and quickly.

Popularity for the term `Cloud-Computing' has been increasing in recent years. There are many great companies such as Yahoo, Google etc. tried to provide related services to business community, even through public users. In addition to the SQL technique, Map-Reduce, a programming model that realizes implementing large-scale data processing, has been a hot topic that is widely discussed through many studies. Many real-world tasks such as data processing for search engines can be parallel-implemented through a simple interface with two functions called Map and Reduce. In this paper, we focus on comparing the performance of the Hadoop implementation of Map-Reduce with SQL Server though simulations. In our studies, Hadoop can complete the same query faster than a SQL Server. On the other hand, some concerned factors are also tested to see whether they would affect the performance for Hadoop or not. We also find that more machines included for data processing can make Hadoop achieve a better performance, especially for a large-scale data set.

Without considering the controversial issue such as copyright, file sharing is the most popular application in a Peer-to-Peer (P2P) network. Piece-based file sharing is an intuitive way to speed up the sharing process in a P2P network, such as the BT P2P network, and Rarest Piece First (RPF) is the most common way to distribute file pieces. As heterogeneous networks have been fast developing, the transmission conditions among nodes therefore become diversified. In the past, few studies had ever been made about how to more efficiently distribute the file pieces based on the transmission conditions among nodes. In the study, we propose two improved algorithms (a) RPF with a Shortest Transmission Time (STT) requesting node first and (b) a STT requesting node first with RPF to solve this problem. The simulation results show that our proposed schemes can perform better than RPF in a heterogeneous network environment.

Thanks to the growing of the wireless networks, the video streaming application becomes a ubiquitous joyful service. In wireless communication networks, the service traffic spans across the wired and wireless domains. Hence, the service of quality (QoS) control becomes complicated. Generally, the QoS is manipulated by the receiving feedback from the mobile user equipments (UEs) to the streaming server. If the feedback latency can be shortened, the streaming service can be adapted according to the individual network condition timely. Meanwhile, a public streaming service is normally realized by multiple uni-casting streams instead of the multicasting ones over IP. To make the service more efficient, an input video session can be encoded as multiple-quality streams so that some UEs with a similar receiving condition can share streams with the same service quality. In this article, SPONGE (Stream Pooler Over a Network Graded Environment) is proposed to improve the adaptive streaming service across wired/wireless networks. SPONGE can alleviate the direct load from the original streaming server to the end UEs and make each UE get an adaptive streaming service according to its network condition timely by the reduced feedback latency of network conditions. Our simulation results show that SPONGE can react to network condition accurately and quickly so as to have a smooth and better playback quality at the end user site across wired/wireless networks.

Peer-to-Peer (P2P) networks have been developed for resource sharing for years. In a P2P network, resource locating is the first and foremost challenge, including issues about long searching time, a large amount of duplicated query message and a small success rate. Many efficient methods have been proposed to resolve these issues. However, most of these proposals may only benefit searching a popular object. In this paper, we propose a Cloning Random Walker with a Dominating Set (CloneRW+DS) method to adaptively locate the target resource no matter the target resource is popular or unpopular in a P2P network. Our simulation results show our proposed scheme can perform a better trade-off among message overhead, searching delay, and success rate while conducting resource locating in a P2P network.

Conventional surveillance systems capture realtime events from cameras mounted at fixed locations. A surveillance user then monitors the occurrences via a browser at a stationary computer. To enhance the mobility at the viewer side and the camera side, we implement a remote mobile surveillance system by integrating some inexpensive techniques - an OSGi service platform, which can easily be implanted with future developed applications, a camera mounted on an embedded system which is carried by a robot - Lego Mind Storms NXT, and a J2ME based viewer and controller program on a mobile phone. With assistance from modern wireless networks, surveillance users can handily monitor the remote events in a wider viewing range at anytime and anywhere.

Compared to the voice service over PSTN, people look forward to a competent Voice over IP (VoIP) service. The introduction of the SIP protocol has a profound impact on the VoIP world. Exploiting more SIP proxy servers can raise the service availability. However, how to assure the efficiency and availability of the service by such an improvement becomes a challenging issue. Based on above, we propose a probing-based name resolution scheme to achieve high availability and load balance for the SIP-based VoIP service. The idea is based on intercepting the prerequisite name resolution process in a typical client-server application in the IP network. We tag the probing mechanism onto the open source project Domain Name Relay Daemon (DNRD) to become a domain name resolution based load balancer (DN-LB). By our scheme, all SIP request messages from the SIP clients can be fairly distributed to a failure-proof SIP proxy server in the server farm without the necessity of using any additional costly intermediate network devices and changing the standard SIP architecture. Therefore, DN-LB can increase the service reliability and fault tolerance with a low cost for the SIP-based VoIP service.

Under the B3G vision, cellular operators construct a complementary environment - public WLAN to make up for the limited data access service to become cellular/PWLAN operators. In addition, confederating various PWLAN operators including PWLAN operators, Cellular/PWLAN operators and PWLAN roaming brokers under the PWLAN roaming agreements can intensify the PWLAN service coverage. To build a better PWLAN access service, the group formed by heterogeneous PWLAN roaming partners is markedly growing. Authentication process in such a cooperative service plays the foremost role for users and conducts lots of back-and-forth authentication messages transmitted among PWLAN partners. To better understand the nuts and bolts of the authentication operation in a cooperative B3G network, this article aims to analyze the authentication process empirically by some experiment results. We believe such an experience and concept can be referred to integrate the future advanced heterogeneous networks.

Wireless sensor networks have been practically applied to many fields including industry, science and environment monitoring. A traditional wireless sensor network is made of several distributed sensors and one sink. The sink gathers and analyzes the data from sensors and then makes some actuator take the corresponding action accordingly. A sensor-sink based architecture is vulnerable to a single-point of failure problem. If a sensor does not have a direct link with the sink, it must deliver data by hopping through other sensors. The sink may take a longer time to collect data and even make a decision. On the other hand, unexpected faulty sensors may result in an incorrect diagnosis. Based on above, we propose an autonomous wireless sensor network where regional sensors determine the local action based on the local measurements to raise the efficiency of action taking and use a consensus problem based algorithm to raise its fault resilience. In this article, we also prove correctness of the proposed scheme and analyze its computation complexity.

Cellular operators offer voice services for years and own a solid amount of users with a certain authority on running cellular business. Recently, various fascinated value added contents and services are emerging in the post-voice era. To extend the business and leverage the exiting operation and maintenance systems, operators strive to become a trusty broker that aggregates a variety of valued added services provided by lots of third-party service/content providers as well as a payment agency that collects the service/content fees into a unified bill for users. In the past, little studies had ever clearly and deeply revealed for how to practically substantiate a lightweight system for connecting users, cellular operators and content/service providers by highly exploiting the mature cellular environment to achieve a triple-win situation. In this article, we elaborate a well- constructed A.A.A. brokering system sitting between the content/service providers and users to link the value chain in such a triple-play game - A4BS, which generalizes certain basic mechanisms that govern authentication, authorization, advice-of- charge issuing, accounting, billing and settlement. Through this tried-and-true system, content/service providers can focus on service creation without much attention to the end-user billing. Meanwhile, users can be charged on a transaction basis with an instant expenditure notification and have a one-stop payment in a single cellular bill as well.

To offer a more efficient wireless data access service than 2G/2.5G/3G networks, public WLAN (PWLAN) stands in a predominant position to embrace the wireless broadband era. Reusing existing mechanisms for user authentication, access control, billing and roaming handling procedures in the mobile territory to construct a complementary network, PWLAN attracts the attention of cellular operators. We investigate a practical end-to-end PWLAN architecture capable of using 2G/3G SIM-based authentication for current mobile users and of simultaneously carrying out Web-based authentication for ordinary users without SIMs (subscriber identity modules). Additionally, we give consideration to confederating various wireless Internet service providers (WISPs) by the RADIUS based roaming mechanism and leverage the existing cellular resource. The proposed considerations and guidelines provide a baseline skeleton for building an extendable and flexible cellular/PWLAN architecture.

A blog (shortened from 'weblog') is a trendy way to share personal journal with others in the cyber world. Traditionally rendering and accessing blogs are normally conducted at a stationary PC. However, such a scheme hinders blog users from writing and reading blogs timely. A short-lived idea came out and passed away suddenly. To facilitate the instant blog updating and retrieving, we combined cellular messaging (SMS/MMS messaging and MMS/WAP push) and open sources (Blosxom and Apache) to develop a novel system for the blog rendering and accessing instantly (BRAINS). BRAINS enables blog users to note down their whims and share interests anytime and anywhere. Blog journalists can utilize their mobile phones in hand to compose and deliver blogs to BRAINS by SMS and MMS messaging at their pleasure. Through p re-registering on BRAINS, readers also can get the up-to-date blogs or notification immediately by MMS push or WAP push respectively. With pervasive networks, BRAINS makes mobile blog rendering and accessing more evident.