The optimal power-flow (OPF) problem, which plays a central role in operating electrical networks, is considered. problem nonconvex and is, fact, NP hard. Therefore, designing efficient algorithms of practical relevance crucial, though their global optimality not guaranteed. Existing semidefinite programming relaxation-based approaches are restricted to OPF problems where zero duality holds. In this paper, an novel method address the general investigated. proposed based on alternating...

Nonconvex and structured optimization problems arise in many engineering applications that demand scalable distributed solution methods. The study of the convergence properties these methods is, general, difficult due to nonconvexity problem. In this paper, two combine fast augmented Lagrangian-based with separability alternating are investigated. first method is adapted from classic quadratic penalty function called direction (ADPM). Unlike original method, where single-step optimizations...

The increased penetration of volatile renewable energy into distribution networks necessities more efficient distributed voltage control. In this paper, we design feedback control algorithms where each bus can inject both active and reactive power the grid to regulate voltages. law on is only based local measurements communication its physical neighbors. Moreover, buses perform their updates asynchronously without receiving information from neighbors for periods time. algorithm enforces hard...

In distributed optimization and machine learning, multiple nodes coordinate to solve large problems. To do this, the need compress important algorithm information bits so that it can be communicated over a digital channel. The communication time of these algorithms follows complex interplay between a) algorithm's convergence properties, b) compression scheme, c) transmission rate offered by We explore relationships for general class linearly convergent algorithms. particular, we illustrate...

Distributed optimization increasingly plays a central role in economical and sustainable operation of cyber-physical systems. Nevertheless, the complete potential technology has not yet been fully exploited practice due to communication limitations posed by real-world infrastructures. This work investigates fundamental properties distributed based on gradient methods, where information is communicated using limited number bits. In particular, general class quantized methods are studied,...

Internet of Underwater Things (IoUT) have gained rapid momentum over the past decade with applications spanning from environmental monitoring and exploration, defence applications, etc. The traditional IoUT systems use machine learning (ML) approaches which cater needs reliability, efficiency timeliness. However, an extensive review various studies conducted highlight significance data privacy security in frameworks as a predominant factor achieving desired outcomes mission critical...

In electricity distribution networks, the increasing penetration of renewable energy generation necessitates faster and more sophisticated voltage controls. Unfortunately, recent research shows that local control fails in achieving desired regulation, unless there is communication between controllers. However, infrastructure for systems less reliable ubiquitous compared to bulk transmission system. this paper, we design distributed uses limited communication. That is, only neighboring buses...

Despite the huge success of reinforcement learning (RL) in solving many difficult problems, its Achilles heel has always been sample inefficiency. On other hand, RL, taking advantage prior knowledge, intentionally or unintentionally, usually avoided, so that, training an agent from scratch is common. This not only causes inefficiency but also endangers safety –especially during exploration. In this paper, we help learn environment by using pre-existing (but necessarily exact complete)...

Parallel stochastic gradient methods are gaining prominence in solving large-scale machine learning problems that involve data distributed across multiple nodes. However, obtaining unbiased gradients, which have been the focus of most theoretical research, is challenging many applications. The estimations easily become biased, for example, when gradients compressed or clipped, shuffled, and meta-learning reinforcement learning. In this work, we establish worst-case bounds on parallel...

With the rise of low-cost launches, miniaturized space technology, and commercialization, cost missions has dropped, leading to a surge in flexible Earth observation satellites. This increased demand for complex diverse imaging products requires addressing multi-objective optimization practice. To this end, we propose agile satellite scheduling problem (MOAEOSSP) model introduce reinforcement learning-based grey wolf (RLMOGWO) algorithm. It aims maximize efficiency while minimizing energy...

Abstract Predicting failures and maintenance time in predictive is challenging due to the scarcity of comprehensive real-world datasets, among those available, few are series format. This paper introduces a real-world, multivariate dataset collected exclusively from single anonymized engine component (Component X) across fleet SCANIA trucks. The includes operational data, repair records, specifications related Component X while maintaining confidentiality through anonymization. It...

Dual decomposition methods are among the most prominent approaches for finding primal/dual saddle point solutions of resource allocation optimization problems. To deploy these in emerging Internet things networks, which will often have limited data rates, it is important to understand communication overhead they require. Motivated by this, we introduce and explore two measures complexity dual identify efficient algorithms. The first measure <inline-formula...

As electric power system operators shift from conventional energy to renewable sources, distribution systems will experience increasing fluctuations in supply. These present the need not only design online decentralized allocation algorithms, but also characterize how effective they are given fast-changing consumer demand and generation. In this paper, we an dual descent (OD3) algorithm determine (in worst case) much of observed social welfare can be explained by generation capacity demand....

In parallel and distributed machine learning multiple nodes or processors coordinate to solve large problems. To do this, the need compress important algorithm information bits so they can communicate. The goal of this paper is explore how we maintain convergence algorithms under such compression. particular, consider a general class linearly convergent parallel/distributed illustrate design quantizers compressing communicated few while still preserving linear convergence. We our results on...

Internet of Underwater Things (IoUT) have gained rapid momentum over the past decade with applications spanning from environmental monitoring and exploration, defence applications, etc. The traditional IoUT systems use machine learning (ML) approaches which cater needs reliability, efficiency timeliness. However, an extensive review various studies conducted highlight significance data privacy security in frameworks as a predominant factor achieving desired outcomes mission critical...

Noisy gradient algorithms have emerged as one of the most popular for distributed optimization with massive data. Choosing proper step-size schedules is an important task to tune in good performance. For attain fast convergence and high accuracy, it intuitive use large step-sizes initial iterations when noise typically small compared algorithm-steps, reduce algorithm progresses. This intuition has been confirmed theory practice stochastic descent. However, similar results are lacking other...

The wide integration of inverter based renewable energy sources (RESs) in modern grids may cause severe voltage violation issues due to high stochastic fluctuations RESs. Existing centralized approaches can achieve optimal results for regulation, but they have communication burdens; existing decentralized methods only require local information, cannot results. Deep reinforcement learning (DRL) are effective deal with uncertainties, it is difficult guarantee secure constraints DRL training....

This paper presents a description of real-world, multivariate time series dataset collected from an anonymized engine component (called Component X) fleet trucks SCANIA, Sweden. includes diverse variables capturing detailed operational data, repair records, and specifications while maintaining confidentiality by anonymization. It is well-suited for range machine learning applications, such as classification, regression, survival analysis, anomaly detection, particularly when applied to...

Distributed control and decision making increasingly play a central role in economical sustainable operation of cyber-physical systems. Nevertheless, the full potential technology has not yet been fully exploited practice due to communication limitations real-world infrastructures. This work investigates fundamental properties gradient methods for distributed optimization, where information is communicated at every iteration, when using limited number bits. In particular, general class...

In electricity distribution networks, the increasing penetration of renewable energy generation necessitates faster and more sophisticated voltage controls. Unfortunately, recent research shows that local control fails in achieving desired regulation, unless there is some communication between controllers. However, infrastructure for systems are less reliable ubiquitous as compared to bulk transmission system. this paper, we design distributed use limited communication. That is, only...

An important task in the Internet of Things (IoT) is field monitoring, where multiple IoT nodes take measurements and communicate them to base station or cloud for processing, inference, analysis. When are high-dimensional (e.g., videos time-series data), networks with limited bandwidth low-power devices may not be able support such frequent transmissions high data rates. To ensure communication efficiency, this article proposes model measurement compression at inference as a deep neural...

The emergence of big data has caused a dramatic shift in the operating regime for optimization algorithms. performance bottleneck, which used to be computations, is now often communications. Several gradient compression techniques have been proposed reduce communication load at price loss solution accuracy. Recently, it shown how errors can compensated algorithm improve Even though convergence guarantees error-compensated algorithms established, there very limited theoretical support...

Distributed resource allocation is a central task in network systems such as smart grids, water distribution networks, and urban transportation systems. When solving problems practice it often important to have non-asymptotic feasibility guarantees for the iterates, since overall-location of resources easily causes break down. In this paper, we develop distributed reallocation algorithm where every iteration produces feasible allocation. The fully sense that nodes communicate only with...