Given a project, the team formation problem (TFP) aims to find a team of experts with all the required skills and minimize the team’s communication cost. The TFP has emerged as a prominent research topic due to its substantial influence on team performance and the subsequent impact on organizational success. Previous research mainly concentrates on proposing heuristic or metaheuristic algorithms, yet fails to consider the historical experience of solving previous TFP. In addition, most existing learning-based approaches rely on complex multistage workflows, rather than learning the entire process end-to-end. This article proposes an end-to-end deep reinforcement learning approach that focuses on effectively reusing previous experience to tackle the TFP. First, we apply the Markov decision process to model the team formation process, in which the state is primarily characterized by the currently formed team and the skill requirements of the project. Second, we propose a self-updating deep reinforcement learning (SURL) model, in which the social network information serves as the input for the static encoding module, while the skills’ experts and the skill requirements of the project act as the inputs for the dynamic encoding module. Next, a novel deep reinforcement learning algorithm is designed to train the model by fully leveraging the historical experience of team formation. Finally, comprehensive experiments with real-world datasets demonstrate that our proposed model outperforms several commonly used optimization algorithms in terms of the average communication cost.

Massive Open Online Courses (MOOCs) belong to a new cloud-based service in education that suffers from low completion rates. Effective pre-learning intervention services, such as recommending courses with a high probability of completion or filtering courses with a very low probability of completion, will encourage students to spend more time and energy on proper courses, thus can reduce the dropout ratio. In practice, intervention services are introduced when students are predicted to drop out. However, existing methods concentrate on analyzing students’ learning actions and predicting final dropout after a period of enrollment, which are insufficient in preventing students from enrolling in unsuitable courses and withdrawing mid-way. This paper presents a neural network-based Explainable Self-supervised Model (ESM) to predict MOOC dropout before enrollment. Specifically, the student's learning actions on an unenrolled course are estimated using previous logs by the neural network. And then, the action's contribution to the completion of a course is calculated in a similar way. Therefore, the probability of completion for an unenrolled course is predicted by aggregating the learning actions and their contribution to the completion. To train the neural network, a self-supervised training strategy is proposed, where enrolled courses in the training data are randomly selected as validation in each epoch. The ESM outperforms existing methods in terms of prediction accuracy and efficiency. The average increment of Area Under the ROC Curve (AUC) and F-score (F1) in the two MOOCs datasets, XuetangX and KDDCUP, are 8.3% and 0.6%, respectively. Furthermore, the two pre-learning intervention services named courses recommendation and courses filtration are proposed. When courses are recommended, the completion rate increased from 22% to 60% in XuetangX, and from 27% to 45% in KDDCUP. By filtering courses predicted with low completion probability, 40% wasted time in uncompleted courses will be saved in XuetangX.

In mobile crowdsensing, numerous smartphone users fulfill a complex environmental or social task in a cooperative way, in which participant recruitment or task allocation is a fundamental issue. Various mechanisms have been proposed to motivate normal users to participate in sensing tasks or provide high-quality sensing data. However, there exist some conflicts among tasks and participants, which make participant recruitment a challenging issue. For this issue, a conflict-aware participant recruitment (CAPR) mechanism is proposed for mobile crowdsensing, where there may exist task correlations and conflicts. First, two definitions of conflicts are introduced, and then, the participant recruitment problem is formalized. Next, an efficient heuristic algorithm is proposed followed by the payment determination and reputation update of participants. Simulation results indicate that the proposed mechanism can effectively improve the platform utility and the average task quality while guaranteeing no conflicts in fulfilling sensing tasks.

Due to the increasing popularity of smart mobile devices, the amount of mobile data communications has led to explosive growth of data traffic in cellular networks. Cellular networks have to face the challenge of huge communication traffic. Offloading data traffic through opportunistic communication among smart mobile devices is a promising solution to partially solve this problem since there is almost no monetary cost for it. Large amount of smart mobile devices can communicate each other using Bluetooth or WIFI Direct in short communication range and they can form an opportunistic mobile social network. The opportunistic communications among smart mobile devices can effectively reduce the amount of cellular data traffic. However, mobile users take a long time to obtain useful data. In order to reduce data communication latency, this paper proposes three data dissemination protocols named RRDP(Request-Reply Dissemination Protocol), RDP(Random Dissemination Protocol) and LDP(LRU Dissemination Protocol) respectively. The three proposed protocols are based on opportunistic sharing policy. Extensive NS-2 simulation results show that (1) on the campus situation, the user's access delay of RDP is 56.4% less than the RRDP and LDP is 44.8% less than RRDP. (2) in the vehicular environment, the user's access delay of RDP is 32.5% less than the RRDP and LDP is 28.1% less than RRDP. RDP is the best protocol.

Biological frequent patterns usually correspond to the important function (or structure) in biological sequences. Along with the rapid growth of biological sequences, it is significant to find frequent patterns over a large bio-sequence efficiently. However, most of existing algorithms need to produce lots of short patterns or projected databases, which influence the efficiency badly and also increase the cost of space. Graphics processing units (GPUs) embracing many core computing devices, have been extensively applied to accelerate computation performance in many areas. In order to meet the demand of biologists, we redefine the frequent pattern problem with length constraints for finding frequent patterns. We present pruning optimization method for the serial algorithm (POSA), and based on this technique, we propose a parallel algorithm (POPA) which not only reduces the time complexity with a low space cost but also obtains better performance on CUDA. To validate the presented algorithms, we implemented the algorithms on multiple-core CPU and various GPU devices. Also, CUDA optimization techniques are applied to speed up calculation in the paper. Finally, experimental results show that compared with the serial algorithm on CPU with six cores, POSA achieves 1.2~4.5 speedup, and POPA gains 3~20 speedup.

The repetitions in biological sequence analysis are of great biological significance. Finding the repetitions has been a hot topic in gene projects naturally. In recent years, graphics processing unit (GPU) has been far exceeded the CPU in terms of computing capability and memory bandwidth, especially CUDA dramatically increases in computing performance by harnessing the power of the GPUs. This paper proposes efficient parallel algorithms on CUDA to accelerate finding PTRs which is redefined as LPRs based on the SUA Index. The proposed parallel algorithms have been utilized with the parallel primitives offered by Thrust library and the effective parallel bit compression technology based on division to achieve better acceleration. Optimization techniques include CUDA streams technology are also realized to reduce transmission latency. Experimental results show that the proposed parallel algorithms are faster than the benchmark with 1.6∼5.4 speedup.

To solve the problems of control channel bottleneck and communication delay in multi-radio wireless sensor networks, this paper designs an asynchronous MAC protocol named UMD2-MAC in which a union mechanism of prime channel hopping(PCH for short) and control-channel reservation is used. The channel hopping makes sure that the average delay of selecting the same channel is small while the reservation at control channel is employed when the time of channel hopping is long. In addition, to reduce ACK overheads, a conflict-notification mechanism is introduced. Through theoretical analysis, this paper observes that the PCH mechanism of UMD2-MAC consumes less time compared to the random channel hopping mechanism. In order to evaluate the performance of UMD2-MAC, extensive simulations are conducted. The simulation results show that UMD2-MAC not only can adapt to the traffic variation, but also outperforms protocols only with random channel hopping and protocols only with control channel reserving in aspects of throughput, energy and latency respectively.

Approximate string matching using the k-difference technique has been widely applied to many fields such as pattern recognition and computational biology. Data dependency exists in the traditional sequential algorithm. Therefore, it is hard to design a parallel algorithm for approximate string matching with k differences. This paper presents a technique to eliminate data dependency. Based on this technique, this paper also presents a parallel algorithm which can calculate the elements in the same row of the edit distance matrix in parallel by eliminating data dependency. The algorithm has high parallelism, but requires synchronization. To validate the proposed algorithm, it is implemented on GPU and multiple-core CPUs. Moreover, the CUDA optimization techniques are also presented in the paper. Finally, experimental results show that, compared with the traditional sequential algorithm on CPU with twenty-four cores, the proposed parallel algorithm achieves speedup of 7-42 on GPU.

Wireless sensor networks(WSNs) are being widely deployed for many monitoring applications, of which a popular one is anti-theft. However, current WSN technologies for anti-theft are either very susceptible to the environment interference or easily compromised by the thieves. Hence, they fail to achieve the desired effectiveness in many anti-theft scenarios. To address these limitations, this paper proposes a novel Hidden Anti-theft System (HAS) to monitor theft intrusion, which is based on the influences of theft intrusion on signal strength according to the shadowing effect in wireless communication. The theft intrusion is detected by finding abnormal RSSI samples of a wireless link compared with the stable range of signal strength in the link's normal state. Through the proposed monitoring approach, HAS can effectively detect intrusion while keeping invisible. In HAS system, to achieve load balance for detection task, we propose an efficient algorithm to determine the set of links that each node monitors. To reduce the response time, a dual-layer scanning solution with dual-radio nodes is proposed to scan the monitoring area more intensively. The experiment results show that the efficiency of HAS. HAS achieves very low false positive and false negative rates, and the response time of dual-layer scanning is 54.2% less than single layer scanning.

We study multi-path reliable routing in wireless sensor networks. When data is transmitting from the source node to the destination node, once forward nodes conflict or failures the data cannot transmit as usual. Data missing can cause a serious impact on the network. Therefore, it is worth investigating the reliability of routing protocols. Aiming at data missing caused by nodes confliction and failures, this paper starts with proposing a novel transfer relationship of two paths by a bridge. It is a notion used to react how the data is transmitted from one path to another path. We then formulate the optimal routing based on the transfer relationship by considering the effect of wireless interference as a bipartite graph model. Through the bipartite graph model, we can calculate the interference between two paths easily. In addition, we propose a strategy to deal with congestion and failures. The real test-bed experimental and simulation results show that the proposed algorithms can deal with network congestion and improve the reliability of transmission significantly.

Solving the eigenvalues of matrices is an open problem which is often related to scientific computation. With the increasing of the order of matrices, traditional sequential algorithms are unable to meet the needs for the calculation time. Although people can use cluster systems in a short time to solve the eigenvalues of large-scale matrices, it will bring an increase in equipment costs and power consumption. This paper proposes a parallel algorithm named Jacobi on gpu which is implemented by CUDA (Computer Unified Device Architecture) on GPU (Graphic Process Unit) to solve the eigenvalues of symmetric matrices. In our experimental environment, we have Intel Core i5-760 quad-core CPU, NVIDIA GeForce GTX460 card, and Win7 64-bit operating system. When the size of matrix is 10240×10240, the number of iterations is 10000 times, the speedup ratio is 13.71. As the size of matrices increase, the speedup ratio increases correspondingly. Moreover, as the number of iterations increases, the speedup ratio is very stable. When the size of matrix is 8192×8192, the number of iterations are 1000, 2000, 4000, 8000 and 16000 respectively, the standard deviation of the speedup ratio is 0.1161. The experimental results show that the Jacobi on gpu algorithm can save more running time than traditional sequential algorithms and the speedup ratio is 3.02~13.71. Therefore, the computing time of traditional sequential algorithms to solve the eigenvalues of matrices is reduced significantly.

Wireless sensors are attached to all kinds of mobile devices/entities such as mobile phones, PDAs, vehicles, robots and animals. This generates Mobile Wireless Sensor Networks (MWSNs) with very dynamic topologies and loose connectivity that depend on mobility of the mobile devices. Data collection from these mobile sensors has become a great challenge considering volatile topologies, loose connectivity and limited buffer storage. This paper proposes a stochastic compressive data collection protocol for MWSNs named SMITE. SMITE consists of three parts: random collector election, stochastic direct transmission from common nodes to collectors when common nodes are in the collectors' transmission range, and angle transmission from collectors to the mobile sink when collectors gather enough data using a predictive method. The collectors use bloom filters to compress the received data. The protocol's performance is theoretically analyzed. The analytic results show that data from the common nodes can be gathered to the collectors with a high probability and gathered data on the collectors can also be forwarded to the mobile sink with a high probability. Simulations are carried out for performance evaluation. The simulation results show that SMITE significantly outperforms the state-of-the-art solutions such as DFT-MSN, SCAR and Sidewinder on the aspects of delivery ratio, transmission overhead, and time delay.

Power control is a complex issue in routing since the increase of transmission power supplies more opportunities to select optimal routes due to the fact that more links are available. However, conversely, it also implies higher interference and hence decreases the performance of routing. Since multi-radio multi-channel schemes can efficiently mitigate interferences through allowing more concurrent transmissions, therefore, it is worth to investigate the routing scheme in Multi-Power Multi-Radio (MPMR) wireless sensor networks (WSNs). In this paper, we study the joint routing, scheduling, channel assignment and power control problem in MPMR WSNs, which is proven a NP-Hard problem. We first formulate the optimal routing problem as a linear programming problem. Subsequently, we develop a distributed routing protocol based on the random walk method which can efficiently decrease the computational complexity in large-scale WSNs by avoiding solving the linear programming problem. Theoretical analysis and simulations show that the routing based on MPMR can improve the data transmission efficiency and the proposed cross-layer routing scheme significantly reduces the energy consumption and the end-to-end transmission delay.

To handle the triple hidden terminal problems, this paper proposes OCO, an asynchronous multi-channel MAC protocol with opportunistic cooperation for wireless sensor networks. By adopting opportunistic cooperation, OCO effectively alleviates, if not eliminates, the triple hidden terminal problems. More importantly, OCO is fully distributed with no requirements of time synchronization or multi-radio scheme, so it is easy to be implemented on the real sensor nodes. Via the theoretical analysis, the opportunistic probability that a node cooperates with its neighbor is obtained. To validate the effectiveness of opportunistic cooperation, extensive simulations and real test bed experiments were con-ducted. The simulation and experimental results show that when the number of channels is large or the network loads are heavy, OCO improves energy efficiency and throughput significantly compared with other works in the literature.

This paper proposes a new Random Channel Selection scheme with probabilistic backoff, called RCS, for a class of multi-channel MAC protocols in heavy loads WSNs to tackle the channel conflict problem. By adopting RCS, a node can reduce the probability of selecting a busy channel for data communication. Therefore, RCS can avoid data packet collision, and thus conserve more energy to extend the lifetime of WSNs. More importantly, RCS is fully distributed with no requirements of time synchronization or multi-radio, so it is practical to realize RCS in resource-constrained sensor nodes. In theoretical analysis, the probability of a channel conflict creation and the average number of misunderstood channels are obtained, which can guide the configurations of RCS. More importantly, RCS is evaluated in both simulation and testbed experiments, and results indicate that as the number of channels and loads increase, RCS significantly improves throughput and energy efficiency as well.

Data aggregation is an essential yet time-consuming task in wireless sensor networks (WSNs). This paper studies the well-known Minimum-Latency Aggregation Schedule (MLAS) problem and proposes an energy-efficient distributed scheduling algorithm named Clu-DDAS based on a novel cluster-based aggregation tree. Our approach differs from all the previous schemes where Connected Dominating Sets or Maximal Independent Sets are employed. We prove that Clu-DDAS has a latency bound of 4R' + 2Delta - 2, where Δ is the maximum degree and R' is the inferior network radius which is smaller than the network radius R. Clu-DDAS has comparable latency as the previously best centralized algorithm E-PAS, while Clu-DDAS consumes 78% less energy as shown by the simulation results. Clu-DDAS outperforms the previously best distributed algorithm DAS whose latency bound is 16R' + Δ - 14 on both latency and energy consumption. On average, Clu-DDAS transmits 67% fewer total messages than DAS does. We also propose an adaptive strategy for updating the schedule to accommodate dynamic network topology.

In this paper, a duty cycle based multi-channel MAC protocol with multiple channel reservation, called M-cube, is proposed to tackle the triple hidden terminal problems. M-cube can make nodes to choose one actually idle channel from all the expected idle channels. Therefore, M-cube can avoid data packet collisions resulted by the triple hidden terminal problems. By minimizing the lower bound of the average number of times of channel switching in M-cube, the optimal duty cycle is obtained through theoretical analysis. To validate the effectiveness of multiple channel reservation and dynamic optimal duty cycling, extensive simulations and real test bed experiments were conducted. Both the simulation and experiment results show that when the number of channels is large or network loads are heavy, M-cube improves energy efficiency and throughput significantly compared with other works in the literature.

This paper proposes ARM, an receiver-initiated MAC protocol with duty cycling to tackle control channel saturation, triple hidden terminal and low broadcast reliability problems in asynchronous multi-channel WSNs. By adopting a receiver-initiated transmission scheme and probability-based random channel selection, ARM effectively solves control channel saturation and triple hidden terminal problems. Further, ARM employs a receiver-adjusted broadcast scheme to guarantee broadcast reliability for broadcast-intensive applications. Via the theoretical analysis, two factors that assist ARM to handle these problems are derived. The simulation and real testbed experimental results show that via solving these three problems ARM achieves significant improvement in energy efficiency and throughput. Moreover, ARM exhibits a prominent ability to enhance its broadcast reliability.

To tackle control channel saturation problems, this paper proposes OMA, an asynchronous duty cycle based multi-channel MAC protocol with opportunistic media access for wireless sensor networks. By adopting opportunistic media access, OMA effectively alleviates, if not completely eliminates, control channel saturation problems. More importantly, OMA is fully distributed with no requirements of time synchronization or multi-radio. Therefore, OMA is very easy to be implemented in resource-constrained sensor nodes. Via the theoretical analysis, the opportunistic probability with which a node opportunistically access the control channel is obtained. To validate the effectiveness of opportunistic media access, extensive simulations and real test bed experiments were conducted. The simulation and experimental results show that when the number of channels is large or the network loads are heavy, OMA improves energy efficiency and throughput significantly compared with other works in the literature.

Data mining in wireless sensor networks (WSNs) is a new emerging research area. This paper investigates the problem of real time clustering of sensory data in WSNs. The objective is to cluster the data collected by sensor nodes in real time according to data similarity in a d-dimensional sensory data space. To perform in-network data clustering efficiently, a Hilbert Curves based mapping algorithm, HilbertMap, is proposed to convert a d-dimensional sensory data space into a two-dimensional area covered by a sensor network. Based on this mapping, a distributed algorithm for clustering sensory data, H-Cluster, is proposed. It guarantees that the communications for sensory data clustering mostly occur among geographically nearby sensor nodes and sensory data clustering is accomplished in in-network manner. Extensive simulation experiments were conducted using both real-world datasets and synthetic datasets to evaluate the algorithms. H-Cluster consistently achieves the lowest data loss rate, the highest energy efficiency, and the best clustering quality.

Area query processing is significant for various applications of wireless sensor networks. No previous study has specifically addressed this issue. We can adopt a naive method, which is to send all data to base station for centralized processing. However, this method wastes a large amount of energy for reporting useless data. This motivates us to propose an energy-efficient in-network area query processing scheme. In our scheme, the whole monitored area is partitioned into grids, and a gray code is used to represent a grid ID (GID), which is a smart way to describe an area. Furthermore, a reporting tree is constructed to process merging areas and aggregations. Based on the properties of GIDs, useless data can be dropped and areas can be merged as early as possible. Incremental update is used to continuously generate query results. In essence, all of these strategies are pivots to conserve energy consumption. With a thorough simulation study, it is shown that our scheme is energy-efficient.

When wireless sensor networks are employed for event monitoring, such as fire detection and enemy movement monitoring, the observers are more interested in the monitored events rather than the readings from sensors. To answer event queries such as "Where was the fire detected during 2-6pm?", an energy-efficient query processing technique is required. This paper presents a data-centric storage strategy, called CM-DCS, and also proposes two distributed event query processing algorithms. Furthermore, the energy consumptions for query processing methods based on three kinds of storage strategies namely external storage, local storage and CM-DCS are analyzed and compared, so that users can have a guideline of choosing a correct storage strategy for different applications. Theoretical analysis and simulation results show that the event query processing algorithm based on CM-DCS can save more energy than those algorithms based on the external storage strategy and the local storage strategy in most cases.