The amount of CO <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$_{2}$</tex-math></inline-formula> emitted per kilowatt-hour on an electricity grid varies by time day and substantially location due to the types generation. Networked collections warehouse scale computers, sometimes called Hyperscale Computing, emit more carbon than needed if operated without regard these variations in intensity. This paper...

Motivated by emerging applications in workforce management, we consider a class of revenue management problems systems with reusable resources. The corresponding are modeled using the well-studied loss network systems. We use an extremely simple linear program (LP) that provides upper bound on best achievable expected long-run rate. optimal solution LP is used to devise conceptually control policy call selection (CSP). Moreover, analyze performance CSP and show it admits uniform guarantees....

Electricity market price predictions enable energy participants to shape their consumption or supply while meeting economic and environmental objectives. By utilizing the basic properties of supply-demand matching process performed by grid operators, known as optimal power flow (OPF), we develop a methodology recover market's structure predict resulting nodal prices using only publicly available data, specifically grid-wide generation type mix, system load, historical prices. Our uses latest...

Bin packing is an algorithmic problem that arises in diverse applications such as remnant inventory systems, shipping logistics, and appointment scheduling. In its simplest variant, a sequence of $T$ items (e.g., orders for raw material, packages delivery) revealed one at time, each item must be packed on arrival available bin pieces material inventory, containers). The sizes are i.i.d. samples from unknown distribution, but the known when arrive. goal to minimize number non-empty bins...

The paper presents accurate and simple dynamic model of a supercapacitor bank system for power dynamics studies. proposed is derived from detailed RC circuit representation. Furthermore, complete control the also presented. easy to integrate in any simulation software consists only up four standard datasheet parameters. performance grid frequency low-voltage ride through illustrated on IEEE 14-bus test DIgSILENT PowerFactory. It shown that case transient stability simulations ideal...

This paper proposes a machine learning-based method for predicting generator rotor angle responses (trajectories) following large disturbance in power system. A Long Short-Term Memory (LSTM)-based Recurrent Neural Network (RNN) is used to predict at any time instant after the fault inception by designing input and output of network with predefined sliding windows. The numbers neurons LSTM Fully-Connected (FC) layers are optimised Particle Swarm Optimisation (PSO) algorithm, which was proved...

Display advertising is the graphical on World Wide Web (WWW) that appears next to content web pages, instant messaging (IM) applications, email, etc. Over past decade, display ads have evolved from simple banner and pop-up include various combinations of text, images, audio, video, animations. As a market segment, continues show substantial growth potential, as evidenced by companies such Microsoft, Yahoo, Google actively vying for share. sales process, are typically sold in packages, result...

Over the past decade, there has been a global growth in datacenter capacity, power consumption and associated costs. Accurate mapping of resource usage (CPU, RAM, etc.) hardware configurations (servers, accelerators, to its is necessary for efficient long-term infrastructure planning real-time compute load management. This paper presents two types statistical models that relate CPU Google's Power Distribution Units (PDUs, commonly referred as domains) their consumption. The are deployed...

We investigate a widely popular least-recently-used (LRU) cache replacement algorithm with semiMarkov modulated requests. SemiMarkov processes provide the flexibility for modeling strong statistical correlation, including broadly reported long-range dependence in World Wide Web page request patterns. When frequency of requesting n is equal to generalized Zipf's law c/n/sup /spl alpha//, alpha/ > 1, our main result shows that fault probability asymptotically, large sizes, same as...

The paper proposes methodology for development of dynamic equivalent model hybrid renewable energy source (HRES) plant suitable reliable assessment the overall transient stability realistic, large power systems. Transient index is used system assessment, and consequently evaluation performance. historical production data transmission network short-circuit fault statistics, together with probabilistically modelled uncertainties associated operation, are to generate a set realistic responses...

We analyse a class of randomized Least Recently Used (LRU) cache replacement algorithms under the independent reference model with generalized Zipf's law request probabilities. The randomization was recently proposed for Web caching as mechanism that discriminates between different document sizes. In particular, maintains an ordered list documents in following way. When size are completely determined by currently requested document. case replacement, necessary number least moved to front...

It was recently proved by Jelenković and Radovanović (2004) that the least-recently-used (LRU) caching policy, in presence of semi-Markov-modulated requests have a generalized Zipf's law popularity distribution, is asymptotically insensitive to correlation request process. However, since previous result asymptotic, it remains unclear how small cache size can become while still retaining preceding insensitivity property. In this paper, assuming are generated nearly completely decomposable...

Abstract Caching is widely recognized as an effective mechanism for improving the performance of World Wide Web. One key components in engineering Web caching systems designing document placement/replacement algorithms updating collection cached documents. The main design objectives such a policy are high cache hit ratio, ease implementation, low complexity and adaptability to fluctuations access patterns. These essentially satisfied by used heuristic called least‐recently‐used (LRU)...

The paper presents the methodology for dynamic equivalent modelling of hybrid renewable energy source (HRES) plant with aim obtaining highly accurate time domain HRES power responses at any during year. focus is on transient stability studies. Historical production dataset and transmission network short-circuit fault statistical data, along unsupervised data mining deep learning techniques, represent a basis procedure. Dynamic model (DEM) developed in form Long Short Term Memory network....

Renewed interest in caching techniques stems from their application to improving the performance of World Wide Web, where storing popular documents proxy caches closer end-users can significantly reduce document download latency and overall network congestion. Rules used update collection frequently accessed inside a cache are referred as replacement algorithms. Due many different factors that influence Web performance, most desirable attributes scheme low complexity high adaptability...

A large number of application areas involve resource allocation problems in which resources different capabilities are used to provide service various classes customers at their arrival instants, otherwise the opportunity serve customer is lost. Stochastic loss networks often capture dynamics and uncertainty this class problems. wide variety examples include applications telephony data networks, distributed computing centers, inventory control manufacturing systems, call contact centers....

We develop a model based on stochastic loss networks to characterize the dynamics and uncertainty in general workforce management optimization. formulate profit maximization problems with serviceability constraints under different assumptions demand supply. Though these optimization are nonlinear programming problems, we able observe some intrinsic properties of functions that facilitate efficient computation optimal solution. Numerical results demonstrate our provides capacity planning...

The analysis of stochastic loss networks has long been interest in computer and communications is becoming important the areas service information systems. In traditional settings computing well-known Erlang formula for blocking probabilities these systems becomes intractable larger resource capacities. Using compound point processes to capture variability request process, we generalize existing models this framework derive simple asymptotic expressions probabilities. addition, extend our...

The advancements in technology have made it possible to automatically record and store large amount of data, which has resulted a need for development application efficient data analysis techniques. Unsupervised clustering methods proven be capable extracting useful information from various types sizes datasets. This paper investigates the performance standard agglomerative hierarchical algorithm using two time series datasets electric power system neuroscience area. main steps procedure are...

Motivated by the problem of packing Virtual Machines on physical servers in cloud, we study online stochastic bin under two settings -- with permanent items, and item departures. In setting present first truly distribution-oblivious heuristic that achieves O(√ n ) regret compared to OPT for all distributions. Our algorithm is essentially gradient descent suitably defined Lagrangian relaxation Linear Program. We also prove guarantees our against non i.i.d. input using a randomly delayed...

short-paper Share on Revenue maximization through "smart" inventory management in reservation-based online advertising Authors: Ana Radovanović Google, Inc., New York, NY, USA USAView Profile , Assaf Zeevi Columbia University, Authors Info & Claims ACM SIGMETRICS Performance Evaluation ReviewVolume 38Issue 2September 2010 pp 33–35https://doi.org/10.1145/1870178.1870190Published:15 October 2010Publication History 3citation137DownloadsMetricsTotal Citations3Total Downloads137Last 12...