Cuirong Wang

Stability-Based RREQ Forwarding Game for Stability-Oriented Route Discovery in MANETs

In traditional stability-oriented route discovery of mobile ad hoc networks, in-between nodes need to rebroadcast identical route request (RREQ) packets, which contain same source node ID and broadcast sequence number, to discover more stable route, yet it increases routing overhead and data transmission delay obviously. Therefore, a stability-oriented route discovery algorithm is proposed to limit routing overhead and decrease transmission delay. In this algorithm, all neighbor nodes of some node will play a mix strategy game named stability-based RREQ forwarding game after receiving an identical RREQ, and independently determine the RREQ forwarding probability based on Nash equilibrium, respectively. The simulation results show that the proposed stability-oriented route discovery algorithm not only reduces routing overhead and transmission delay effectively, but also improve other routing performance.

DLRDG: distributed linear regression-based hierarchical data gathering framework in wireless sensor network

For many applications in wireless sensor network (WSN), the gathering of the holistic sensor measurements is difficult due to stringent constraint on network resources, frequent link, indeterminate variations in sensor readings, and node failures. As such, sensory data extraction and prediction technique emerge to exploit the spatio-temporal correlation of measurements and represent samples of the true state of the monitoring area at a minimal communication cost. In this paper, we present DLRDG strategy, a distributed linear regression-based data gathering framework in clustered WSNs. The framework can realize the approximate representation of original sensory data by less than a prespecified threshold while significantly reducing the communication energy requirements. Cluster-head (CH) nodes in WSN maintain linear regression model and use historical sensory data to perform estimation of the actual monitoring measurements. Rather than transmitting original measurements to sink node, CH nodes communicate constraints on the model parameters. Relying on the linear regression model, we improved the CH node function of representative EADEEG (an energy-aware data gathering protocol for WSNs) protocol for estimating the energy consumption of the proposed strategy, under specific settings. The theoretical analysis and experimental results show that the proposed framework can implement sensory data prediction and extracting with tolerable error bound. Furthermore, the designed framework can achieve more energy savings than other schemes and maintain the satisfactory fault identification rate on case of occurrence of the mutation sensor readings.

An Integrated Framework for Analysis and Mining of the Massive Sensor Data Using Feature Preserving Strategy on Cloud Computing

Cloud computing can provide a powerful, scalable storage and the massive data processing infrastructure to perform both online and offline analysis and mining of the heterogeneous sensor data streams. In contrast to traditional data objects, the sensor data objects from the Internet of Thing (IoT) monitoring application have continuously changing, high-dimensional, spatiotemporal relation and heterogeneous attributes. Therefore, the analysis and mining problem of the massive sensor data objects can be more complicated. The paper formally presents an integrated framework for analysis problem of the massive sensor data with insights into the high-dimensional problem using the feature preserving on cloud computing. The proposed framework realized the cloud resources independent dynamic allocation and scheduling for the massive sensor data mining using kernel methods for reducing the computation of spatial data retrieval. As the experiment results shown, the strategy can preserve important spatial feature information and provide effective preprocessing analysis results.

Repeatable optimization algorithm based discrete PSO for virtual network embedding

Aiming at reducing the link load and improving substrate network resource utilization ratio, we model the virtual network embedding (VNE) problem as an integer linear programming and present a discrete particle swarm optimization based algorithm to solve the problem. The approach allows multiple virtual nodes of the same VN can be embedded into the same physical node as long as there is enough resource capacity. It not only can cut down embedding processes of virtual link and reduce the embedding time, but also can save the physical link cost and make more virtual networks to be embedded at the same time. Simulation results demonstrate that comparing with the existing VNE algorithm, the proposed algorithm performs better for accessing more virtual networks and reducing embedding cost.

Utility-Driven share scheduling algorithm in hadoop

Job scheduling in hadoop is a hot topic, however, current research mainly focuses on the time optimization in scheduling. With the trend of providing hadoop as a service to the public or specified groups, more factors should be considered, such as time and cost. To solve this problem, we present a utility-driven share scheduling algorithm. Considering time and cost, algorithm offers a global optimization scheduling scheme according to the workload of the job. Furthermore, we present a model that can estimate job execute time by cost. Finally, we implement the algorithm and experiment it in a hadoop cluster.

Fault Tolerant Virtual Network Embedding Algorithm Based on Redundant Backup Resource

Network virtualization allows multiple virtual networks to coexist on a shared physical substrate infrastructure. As network virtualization becomes popular, the problem of efficiently mapping a virtual network while guaranteeing its survivability in the event of failures becomes increasingly important. In this paper, the reliability problem of virtual network is solved based on a node and link redundant backup strategy. We model the virtual network mapping problem as an integer linear programming and present a discrete particle swarm optimization based algorithm to solve the problem. Experimental results show that the proposed algorithm has higher recovery success ratio and can reduce backup bandwidth simultaneously.

An integrated framework for managing massive and heterogeneous sensor data using cloud computing

Cloud computing can provide a powerful, scalable storage and the massive data processing infrastructure to perform both online and offline analysis and mining of the heterogeneous sensor data streams. With the recent explosion of wireless sensor networks and their applicability in military and civilian applications, there is an emerging vision for integrating sensor networks into the cloud computing platform. In contrast to traditional data objects, the sensor data objects have continuously changed, high-dimensional, spatiotemporal relation and heterogeneous attributes. Therefore, the management and processing problem of the massive sensor data objects can be more complicated. The paper formally presents an integrated framework for managing massive and heterogeneous sensor data with insights into the high-dimensional problem using the map-reduce platform of cloud computing. The proposed framework incorporates key concepts such as parallel-processing, scalability and flexibility of resources, sensor data uncertainty and the dynamic deployment and management of applications.

Discrete Particle Swarm Optimization Algorithm for Virtual Network Reconfiguration

Network virtualization allows multiple virtual networks (VNs) to coexist on a shared physical substrate infrastructure. Efficient network resource utilization is crucial for such problem. Most of the current researches focus on algorithms to allocate resources to VNs in mapping. However, reconfiguration problem of running VNs is relatively less explored. Aiming at dynamic scheduling of running VNs, this paper introduces a virtual network reconfiguration model to achieve more substrate network resource utilization. We formulate the virtual network reconfiguration problem as a multi object optimal problem and use discrete particle swarm optimization (DPSO) algorithm to search optimal solution. Experimental results show that by rescheduling the running VNs on substrate network according to the optimal reconfiguration solution our approach can observably reduce the biggest load in both physical node and link load, balance average load and avoid bottlenecks in substrate network so as to gain high VNs accept ratio.

Virtual network embedding algorithm based connective degree and comprehensive capacity

Network virtualization allows multiple virtual networks to coexist on a shared physical substrate infrastructure. As far as possible to support more virtual networks on a shared substrate network, efficient physical resource utilization is crucial. This paper presents a novel approach to increase utility of the substrate network. As an optimization problem, such virtual network embedding problem is formulated as an integer linear programming model. By introducing a sliding window with priority for VN requests, the algorithm embeds virtual nodes based on both connective degree and comprehensive capacity. Experimental results show that comparing with the existing VNE algorithm the proposed algorithm achieve higher VNs accept ratio and gain higher revenue-cost ratio for substrate network.

A Novel Algorithm for Finding Overlapping Communities in Networks Based on Label Propagation

Community discovery in Social network is one of the hot spots. In real networks, some nodes belong to several different communities. Overlapping community discovery has been more and more popular. Label propagation algorithm has been proven to be an effective method for complex network community discovery, this algorithm has the characteristics of simple and fast. For the poor stability problem of Label propagation algorithm, this article proposes a stable overlapping communities discovery method based on the label propagation algorithm: SALPA. At the beginning of the method, introduce the influence of nodes, which is used to measure the influence of nodes, select the most influential nodes as the core nodes, in the propagating stage, when there are more than one label with the same degree of membership, select the connectivity lager than the threshold. The method has been carried out in three real networks and two big synthetic networks. Compared with the classical algorithm, experiment results demonstrate the effectiveness, stability and computational speed of the method have been improved.