Robust Distributed Algorithm Research Articles

Customized decentralized autonomous organization based optimal energy management for smart buildings

In recent years, the unprecedented rising urbanization level and proliferation of distributed energy sources (DES) are motivating the transition of smart building from consumer to prosumer. In the context of mitigating upstream emission via reducing reliance on utility grids, the key challenge becomes realizing autonomous and economically optimal energy management within smart buildings. This paper innovatively proposes a Decentralized Autonomous Organization (DAO)-based energy management framework for smart buildings, where the mechanism of day-ahead scheduling DAO considering peer-to-peer (P2P) energy sharing is specifically investigated. Besides, a sequential preference satisfaction mechanism is designed to customize participants’ special demands, thereby achieving effective energy matching. Considering potential communication failures between buildings in practical operations, this study develops a communication failure robust distributed algorithm to minimize operational costs and safeguard building’s privacy. Extensive numerical studies for highly DES penetrated smart buildings successfully offers an operation cost reduction of 33.7% when considering preference satisfaction and P2P energy sharing in the day-ahead scheduling problem. Simulation results also demonstrates that the algorithm in this work excels in resisting potential communication failure during operation.

Distributed adaptive Huber regression

Distributed data naturally arise in scenarios involving multiple sources of observations, each stored at a different location. Directly pooling all the data together is often prohibited due to limited bandwidth and storage, or due to privacy protocols. A new robust distributed algorithm is introduced for fitting linear regressions when data are subject to heavy-tailed and/or asymmetric errors with finite second moments. The algorithm only communicates gradient information at each iteration, and therefore is communication-efficient. To achieve the bias-robustness tradeoff, the key is a novel double-robustification approach that applies on both the local and global objective functions. Statistically, the resulting estimator achieves the centralized nonasymptotic error bound as if all the data were pooled together and came from a distribution with sub-Gaussian tails. Under a finite (2+δ)-th moment condition, a Berry-Esseen bound for the distributed estimator is established, based on which robust confidence intervals are constructed. In high dimensions, the proposed doubly-robustified loss function is complemented with ℓ1-penalization for fitting sparse linear models with distributed data. Numerical studies further confirm that compared with extant distributed methods, the proposed methods achieve near-optimal accuracy with low variability and better coverage with tighter confidence width.

Energy efficiency maximization oriented resource allocation in 5G ultra-dense network: Centralized and distributed algorithms

Spurred by both economic and environmental concerns, energy efficiency (EE) has now become one of the key pillars for the fifth generation (5G) mobile communication networks. To maximize the downlink EE of the 5G ultra dense network (UDN), we formulate a constrained EE maximization problem and translate it into a convex representation based on the fractional programming theory. To solve this problem, we first adopt a centralized algorithm to reach the optimum based on Dinkelbach’s procedure. To improve the efficiency and reduce the computational complexity, we further propose a distributed iteration resource allocation algorithm based on alternating direction method of multipliers (ADMM). For the proposed distributed algorithm, the local and dual variables are updated by each base station (BS) in parallel and independently, and the global variables are updated through the coordination and information exchange among BSs. Moreover, as the noise may lead to imperfect information exchange among BSs, the global variables update may be subject to failure. To cope with this problem, we propose a robust distributed algorithm, for which the global variable only updates as the information exchange is successful. We prove that this modified robust distributed algorithm converges to the optimal solution of the primal problem almost surely. Simulation results validate our proposed centralized and distributed algorithms. Especially, the proposed robust distributed algorithm can effectively eliminate the impact of noise and converge to the optimal value at the cost of a little increase of computational complexity.

Iterative Learning Control Based Robust Distributed Algorithm for Non-Holonomic Mobile Robots Formation

This paper presents an iterative learning control-based robust distributed algorithm on the formation issue for a group of differential-drive mobile robots. The fundamental robustness problem in practical application involving initial state shifts, disturbances, noises, and communication time-delays are considered. The distributed algorithm is proposed for each robot in directed network, which is based on the iterative learning rule with both predictive and current learning terms. It is shown that the convergence of formation tracking objective can be guaranteed under a matrix norm condition by using the two-dimensional analysis approach. Numerical simulation and experiment are both given to validate the effectiveness of the proposed algorithm.

Game theory in wireless and communication networks: theory, models, and applications

This unified treatment of game theory focuses on finding state-of-the-art solutions to issues surrounding the next generation of wireless and communications networks. Future networks will rely on autonomous and distributed architectures to improve the efficiency and flexibility of mobile applications, and game theory provides the ideal framework for designing efficient and robust distributed algorithms. This book enables readers to develop a solid understanding of game theory, its applications and its use as an effective tool for addressing wireless communication and networking problems. The key results and tools of game theory are covered, as are various real-world technologies including 3G networks, wireless LANs, sensor networks, dynamic spectrum access and cognitive networks. The book also covers a wide range of techniques for modeling, designing and analysing communication networks using game theory, as well as state-of-the-art distributed design techniques. This is an ideal resource for communications engineers, researchers, and graduate and undergraduate students.

Distributed Decision‐Tree Induction in Peer‐to‐Peer Systems

AbstractThis paper offers a scalable and robust distributed algorithm for decision‐tree induction in large peer‐to‐peer (P2P) environments. Computing a decision tree in such large distributed systems using standard centralized algorithms can be very communication‐expensive and impractical because of the synchronization requirements. The problem becomes even more challenging in the distributed stream monitoring scenario where the decision tree needs to be updated in response to changes in the data distribution. This paper presents an alternate solution that works in a completely asynchronous manner in distributed environments and offers low communication overhead, a necessity for scalability. It also seamlessly handles changes in data and peer failures. The paper presents extensive experimental results to corroborate the theoretical claims. Copyright © 2008 Wiley Periodicals, Inc., A Wiley Company Statistical Analy Data Mining 1: 000‐000, 2008