Wireless sensor networks are often densely deployed for environmental monitoring applications. Collecting raw data from these networks can lead to excessive energy consumption. Thus using the spatial and temporal corr...
详细信息
ISBN:
(纸本)9783540730897
Wireless sensor networks are often densely deployed for environmental monitoring applications. Collecting raw data from these networks can lead to excessive energy consumption. Thus using the spatial and temporal correlations that exist between adjacent nodes we appoint a few as representative nodes that perform in-network aggregation. This reduces the total number of transmissions. Our distributed scheduling algorithm autonomously assigns a particular node to perform aggregation and reassigns schedules when network topology changes. These topology changes are detected using cross-layer information from the underlying MAC layer. We also present theoretical performance estimates and upper bounds of our algorithm and evaluate it by implementing the algorithm on actual sensor nodes, demonstrating an energy-saving of up to 80% compared to raw data collection.
distributed scalable video coding (DSVC) has recently been gaining many attentions due to its benefits in terms of computational complexity, error resilience and scalability, which are important for emerging video app...
详细信息
ISBN:
(纸本)9781538679630
distributed scalable video coding (DSVC) has recently been gaining many attentions due to its benefits in terms of computational complexity, error resilience and scalability, which are important for emerging video applications like wireless sensor networks and visual surveillance system (VSS). In DSVC, the side information (SI) creation plays a key role as it directly affects the DSVC compression performance and the encoder/decoder computational complexity. However, for many VSS applications, the energy of each VSS node is usually attenuating along the time, making the difficulty in transmitting surveillance video in real time. To address this problem, we propose a complexity controlled SI creation solution for the newly DSVC framework. To achieve the flexible SI creation, the complexity associated with SI creation process is modeled using a linear model in which the model parameters are estimated from a fitting process. To adjust the SI complexity, a user parameter is defined based on the availability of the VSS energy resource. Experiments conducted for a rich set of video surveillance data have revealed the benefits of the proposed complexity control solution, notably in both complexity control and compression performance.
Environmental sensor Networks (ESNs) in forests facilitate the study of fundamental processes, and the development of wireless communication makes ESNs into 'intelligent' sensor network, named as Wireless Envi...
详细信息
ISBN:
(纸本)9781509062966
Environmental sensor Networks (ESNs) in forests facilitate the study of fundamental processes, and the development of wireless communication makes ESNs into 'intelligent' sensor network, named as Wireless Environmental sensor Networks (WESNs). However, data loss is prevalent in wireless transmission, which may result in incompletion of sensory datasets. Thus, if we want to achieve a satisfactory accuracy in a WESNs system, the task of recovering data from achieved sensory datasets is unavoidable. Previous works provide many approaches to solve the data missing problem. Compared with other methods, Compressing Sensing (CS) is powerful technique for estimating data, which can utilize a small fraction of data to reconstruct the entire dataset. In real forests, because of the complicated geographic conditions and deployment of sensors, sensory data will largely loss during the wireless transmission. Despite CS technique is a better choice, it cannot be directly applied for the data missing problem. In this paper, we will present a reliable WESNs system and a better approach based on CS to reconstruct sensory datasets.
distributed deep learning systems place stringent requirement on communication bandwidth in its model training with large volumes of input data under user-time constraint. The communications take place mainly between ...
详细信息
ISBN:
(纸本)9781538625880
distributed deep learning systems place stringent requirement on communication bandwidth in its model training with large volumes of input data under user-time constraint. The communications take place mainly between cluster of worker nodes for training data and parameter servers for maintaining a global trained model. For fast convergence the worker nodes and parameter servers have to frequently exchange billions of parameters to quickly broadcast updates and minimize staleness. Demand on the bandwidth becomes even higher with the introduction of dedicated GPUs in the computation. While rdMA-capable network has a great potential to provide sufficiently high bandwidth, its current use over TCP/IP or tied to particular programming models, such as MPI, limits its capability to break the bandwidth bottleneck. In this work we propose irdMA, an rdMA-based parameter server architecture optimized for high-performance network environment supporting both GPU- and CPU-based training. It utilizes native asynchronous rdMA verbs to achieve network line speed while minimizing the communication processing cost on both worker and parameter-server sides. Furthermore, irdMA exposes the parameter server system as a POSIX-compatible file API for convenient support of load balance and fault tolerance as well as its easy use. We have implemented irdMA at IBM's deep learning platform. Experiment results show that our design can help deep learning applications, including image recognition and language classification, to achieve near-linear improvement on convergence speed and training accuracy acceleration by using distributedcomputing resources. From the system perspective, irdMA can efficiently utilize about 95% network bandwidth of fast networks to synchronize models among distributed training processes.
Newly arising IoT-driven use cases often require low-latency analytics to derive time-sensitive actions, where a centralized cloud approach is not applicable. An emerging computing paradigm, referred to as fog computi...
详细信息
ISBN:
(纸本)9781728123653
Newly arising IoT-driven use cases often require low-latency analytics to derive time-sensitive actions, where a centralized cloud approach is not applicable. An emerging computing paradigm, referred to as fog computing, shifts the focus away from the central cloud by offloading specific computational parts of analytical stream processing pipelines (SPP) towards the edge of the network, thus leveraging existing resources close to where data is generated. However, in scenarios of mobile edge nodes, the inherent context changes need to be incorporated in the underlying fog cluster management, thus accounting for the dynamics by relocating certain processing elements of these SPP. This paper presents our initial work on a conceptual architecture for context-aware and dynamic management of SPP in the fog. We provide preliminary results, showing the general feasibility of relocating processing elements according to changes in the geolocation.
Multi-sensor tracking fusion plays a fundamental role in networked information system, especially in the field of fire control systems. According to the diversity, networked and flexible recombined characteristics of ...
详细信息
ISBN:
(纸本)9781479906109;9781479906123
Multi-sensor tracking fusion plays a fundamental role in networked information system, especially in the field of fire control systems. According to the diversity, networked and flexible recombined characteristics of the modern information system, a bottom-up architecture of networked information system and the method of track fusion are investigated. distributed track fusion problem under limited communication is discussed, and federated Kalman consensus filtering(FKCF) algorithm is proposed. Compared to conventional federated filter, FKCF algorithm considers the mobile sensor model, applies Kalman consensus filter to design the sub-filter and designs information-driven method to improve information allocation. The algorithm not only achieves auto recombination and improves survivability, but increases fused tracking accuracy of mobile sensor network with limited communication capability. The experimental results show that FKCF algorithm is better than conventional federated filtering algorithm in track fusion with limited communication.
Energy efficiency is main design issue for protocols of wireless sensor networks. Node clustering is an energy efficient approach for sensor networks. In clustering algorithms, nodes are grouped into independent clust...
详细信息
ISBN:
(纸本)9781467345286
Energy efficiency is main design issue for protocols of wireless sensor networks. Node clustering is an energy efficient approach for sensor networks. In clustering algorithms, nodes are grouped into independent clusters and each cluster has a cluster head. Data units gathered at base station depends upon lifetime of network. Cluster head selection is an important issue for energy efficiency of clustering schemes. Intra cluster communication distance depends upon position of cluster head in cluster. In this paper, a new cluster head selection scheme is proposed. Proposed scheme can be implemented with any distributed clustering scheme. In proposed scheme, network area is divided into two parts: border area and inner area. Scheme restricts cluster head selection to only inner area nodes. Scheme is implemented and simulated with LEACH in NS-2. Simulation shows that proposed scheme significantly outperform LEACH for network lifetime and data gathering rate.
暂无评论