The Experts below are selected from a list of 54003 Experts worldwide ranked by ideXlab platform
Ivan Stojmenovic - One of the best experts on this subject based on the ideXlab platform.
-
secure and energy efficient data aggregation with malicious aggregator identification in wireless sensor networks
International Conference on Algorithms and Architectures for Parallel Processing, 2011Co-Authors: Ivan StojmenovicAbstract:Data aggregation in wireless sensor networks is employed to reduce the Communication Overhead and prolong the network lifetime. However, an adversary may compromise some sensor nodes, and use them to forge false values as the aggregation result. Previous secure data aggregation schemes have tackled this problem from different angles. The goal of those algorithms is to ensure that the Base Station (BS) does not accept any forged aggregation results. But none of them have tried to detect the nodes that inject into the network bogus aggregation results. Moreover, most of them usually have a Communication Overhead that is (at best) logarithmic per node. In this paper, we propose a secure and energy-efficient data aggregation scheme that can detect the malicious nodes with a constant per node Communication Overhead. In our solution, all aggregation results are signed with the private keys of the aggregators so that they cannot be altered by others. Nodes on each link additionally use their pairwise shared key for secure Communications. Each node receives the aggregation results from its parent (sent by the parent of its parent) and its siblings (via its parent node), and verifies the aggregation result of the parent node. Theoretical analysis on energy consumption and Communication Overhead accords with our comparison based simulation study over random data aggregation trees.
-
localized sensor area coverage with low Communication Overhead
IEEE Transactions on Mobile Computing, 2008Co-Authors: Antoine Gallais, Jean Carle, David Simplotryl, Ivan StojmenovicAbstract:We propose several localized sensor area coverage protocols for heterogeneous sensors, each with arbitrary sensing and transmission radii. The approach has a very small Communication Overhead since prior knowledge about neighbor existence is not required. Each node selects a random time out and listens to messages sent by other nodes before the time out expires. Sensor nodes whose sensing area is not fully covered (or fully covered but with a disconnected set of active sensors) when the deadline expires decide to remain active for the considered round and transmit an activity message announcing it. There are four variants in our approach, depending on whether or not withdrawal and retreat messages are transmitted. Covered nodes decide to sleep, with or without transmitting a withdrawal message to inform neighbors about the status. After hearing from more neighbors, active sensors may observe that they became covered and may decide to alter their original decision and transmit a retreat message. Our simulations show a largely reduced message Overhead while preserving coverage quality for the ideal MAC/physical layer. Compared to an existing method (based on hello messages followed by retreat ones and where excessive message loss contributed to excessive coverage holes), our approach has shown robustness in a model with collisions and/or a realistic physical layer.
-
localized sensor area coverage with low Communication Overhead
IEEE International Conference on Pervasive Computing and Communications, 2006Co-Authors: Antoine Gallais, Jean Carle, David Simplotryl, Ivan StojmenovicAbstract:We propose several localized sensor area coverage protocols, for arbitrary ratio of sensing and transmission radii. Sensors are assumed to be time synchronized, and active sensors are determined at the beginning of each round. The approach has a very small Communication Overhead since prior knowledge about neighbor existence is not required. Each node selects a random timeout and listens to messages sent by other nodes before the timeout expires. Sensor nodes whose sensing area is not fully covered (or fully covered but with a disconnected set of active sensors) when the deadline expires decide to remain active for the considered round, and transmit a message announcing their activity status. There are four variants in our approach, depending on whether or not negative and retreat messages are transmitted. Experimental results with ideal MAC layer show that, for a similar number of selected active sensors, our methods significantly reduce number of messages to decide activity compared to existing localized protocol, we also consider a MAC layer with collisions, and show that existing compared method, for dense networks, fails to cover the area reasonably. Our methods, however, still remain robust in terms of high area coverage with reasonable amount of active nodes, despite some message collisions.
Kannan Ramchandran - One of the best experts on this subject based on the ideXlab platform.
-
robust class parallelism error resilient parallel inference with low Communication cost
Asilomar Conference on Signals Systems and Computers, 2020Co-Authors: Yaoqing Yang, Jichan Chung, Guanhua Wang, Vipul Gupta, Adarsh Karnati, Kenan Jiang, Ion Stoica, Joseph E Gonzalez, Kannan RamchandranAbstract:Model parallelism is a standard paradigm to decouple a deep neural network (DNN) into sub-nets when the model is large. Recent advances in class parallelism significantly reduce the Communication Overhead of model parallelism to a single floating-point number per machine per iteration. However, traditional fault-tolerance schemes, when applied to class parallelism, require storing the entire model on the hard disk. Thus, these schemes are not suitable for soft and frequent system noise such as stragglers (temporarily slow worker machines). In this paper, we propose an erasure-coding based redundant computing technique called robust class parallelism to improve the error resilience of model parallelism. We show that by introducing slight Overhead in the computation at each machine, we can obtain robustness to soft system noise while maintaining the low Communication Overhead in class parallelism. More importantly, we show that on standard classification tasks, robust class parallelism maintains the state-of-the-art performance.
-
Communication efficient gradient coding for straggler mitigation in distributed learning
International Symposium on Information Theory, 2020Co-Authors: Swanand Kadhe, Ozan O Koyluoglu, Kannan RamchandranAbstract:Distributed implementations of gradient-based methods, wherein a server distributes gradient computations across worker machines, need to overcome two limitations: delays caused by slow running machines called stragglers, and Communication Overheads. Recently, Ye and Abbe [ICML 2018] proposed a coding-theoretic paradigm to characterize a fundamental trade-off between computation load per worker, Communication Overhead per worker, and straggler tolerance. However, their proposed coding schemes suffer from heavy decoding complexity and poor numerical stability. In this paper, we develop a Communication-efficient gradient coding framework to overcome these drawbacks. Our proposed framework enables using any linear code to design the encoding and decoding functions. When a particular code is used in this framework, its block-length determines the computation load, dimension determines the Communication Overhead, and minimum distance determines the straggler tolerance. The flexibility of choosing a code allows us to gracefully trade-off the straggler threshold and Communication Overhead for smaller decoding complexity and higher numerical stability. Further, we show that using a maximum distance separable (MDS) code generated by a random Gaussian matrix in our framework yields a gradient code that is optimal with respect to the trade-off and, in addition, satisfies stronger guarantees on numerical stability as compared to the previously proposed schemes. Finally, we evaluate our proposed framework on Amazon EC2 and demonstrate that it reduces the average iteration time by 16% as compared to prior gradient coding schemes.
Siguang Chen - One of the best experts on this subject based on the ideXlab platform.
-
Edge Blockchain Assisted Lightweight Privacy-preserving Data Aggregation for Smart Grid
IEEE Transactions on Network and Service Management, 2021Co-Authors: Zhihao Ren, Siguang ChenAbstract:Compared with traditional power systems, smart grid is designed to provide effective and secure energy services. Data aggregation is one of the key technologies in wireless sensor networks, which reduces the amount of data transmission between nodes by merging similar data and simplifying redundant data, thus significantly reducing the computation cost and Communication Overhead of the system. Many data aggregation schemes have been developed for the smart grid in the past years. However, most of the data aggregation schemes ignore the data security and privacy protection issues of the edge layer. To solve these problems, in this paper, we propose an edge blockchain assisted lightweight privacy-preserving data aggregation for smart grid, named EBDA. In this work, we integrate edge computing and blockchain to design a three-layer architecture data aggregation scheme for smart grid. This new architecture supports a two-level data aggregation scheme, which is more efficient and secure. Through theoretical analysis and simulations, EBDA shows great superiority in terms of resisting network attacks, reducing system computation costs and Communication Overhead compared with existing schemes.
Antoine Gallais - One of the best experts on this subject based on the ideXlab platform.
-
localized sensor area coverage with low Communication Overhead
IEEE Transactions on Mobile Computing, 2008Co-Authors: Antoine Gallais, Jean Carle, David Simplotryl, Ivan StojmenovicAbstract:We propose several localized sensor area coverage protocols for heterogeneous sensors, each with arbitrary sensing and transmission radii. The approach has a very small Communication Overhead since prior knowledge about neighbor existence is not required. Each node selects a random time out and listens to messages sent by other nodes before the time out expires. Sensor nodes whose sensing area is not fully covered (or fully covered but with a disconnected set of active sensors) when the deadline expires decide to remain active for the considered round and transmit an activity message announcing it. There are four variants in our approach, depending on whether or not withdrawal and retreat messages are transmitted. Covered nodes decide to sleep, with or without transmitting a withdrawal message to inform neighbors about the status. After hearing from more neighbors, active sensors may observe that they became covered and may decide to alter their original decision and transmit a retreat message. Our simulations show a largely reduced message Overhead while preserving coverage quality for the ideal MAC/physical layer. Compared to an existing method (based on hello messages followed by retreat ones and where excessive message loss contributed to excessive coverage holes), our approach has shown robustness in a model with collisions and/or a realistic physical layer.
-
localized sensor area coverage with low Communication Overhead
IEEE International Conference on Pervasive Computing and Communications, 2006Co-Authors: Antoine Gallais, Jean Carle, David Simplotryl, Ivan StojmenovicAbstract:We propose several localized sensor area coverage protocols, for arbitrary ratio of sensing and transmission radii. Sensors are assumed to be time synchronized, and active sensors are determined at the beginning of each round. The approach has a very small Communication Overhead since prior knowledge about neighbor existence is not required. Each node selects a random timeout and listens to messages sent by other nodes before the timeout expires. Sensor nodes whose sensing area is not fully covered (or fully covered but with a disconnected set of active sensors) when the deadline expires decide to remain active for the considered round, and transmit a message announcing their activity status. There are four variants in our approach, depending on whether or not negative and retreat messages are transmitted. Experimental results with ideal MAC layer show that, for a similar number of selected active sensors, our methods significantly reduce number of messages to decide activity compared to existing localized protocol, we also consider a MAC layer with collisions, and show that existing compared method, for dense networks, fails to cover the area reasonably. Our methods, however, still remain robust in terms of high area coverage with reasonable amount of active nodes, despite some message collisions.
Hiroshi Tsuji - One of the best experts on this subject based on the ideXlab platform.
-
server driven type Communication protocol with data segmentation for remote database access
Electrical Engineering in Japan, 2007Co-Authors: Hiroshi Yamamoto, Hiroshi TsujiAbstract:This paper presents the Communication protocol between applications and remote databases in enterprise systems. Although the International Standard OSI-RDA adopted the one request–one response Communication protocol for the remote database access, it is not suitable from the viewpoints of system resources occupation and Communication Overhead in the case where the system should handle the large volumes of multimedia data. This paper proposes a server-driven Communication protocol in order to solve these problems, and compares the traditional method and the proposed method as to (1) the number of Communications, (2) the number of encoding–decoding, and (3) the dynamic steps. According to the quantitative comparison, the proposed method makes an effort to avoid Communication resources occupation and improves by 50% the Communication Overhead in the case of large Communication data. This paper also suggests a guideline for suitable number of data segmentations for the remote database access. © 2007 Wiley Periodicals, Inc. Electr Eng Jpn, 159(2): 64–71, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.20139
-
server driven type Communication protocol with data segmentation for remote database access
Ieej Transactions on Electronics Information and Systems, 2004Co-Authors: Hiroshi Yamamoto, Hiroshi TsujiAbstract:This paper presents the Communication protocol between applications and remote databases in enterprise systems. Although the International Standard OSI-RDA has presupposed of the 1 request - 1 response Communication protocol for the remote database access, it is not suitable from the view of system resources occupation and Communication Overhead in the case that the system should handle the large volumes of multimedia data. This paper proposes a server driven Communication protocol in order to solve these problems, and compares the traditional method and the proposed method on the followings: (1) the number of Communications, (2) the number of the encoding-decoding, and (3) the dynamic steps. According to the quantitative comparison, the proposed method makes the effort of avoiding Communication resources occupation and improves 50% of Communication Overhead in case of only a large Communication data. This paper also implies the guideline for suitable number of data segmentations for the remote database access.
