arXiv subjects

David Greenwood

Publications and source records attributed to David Greenwood.

16 recordsLinked to original sources

arXiv2026

Improving the Energy Efficiency of High Throughput Computing: A Measurement-Based Case Study

The significant energy consumed by data centers has become a concern both for costs and associated carbon emissions. In particular, the energy efficiency of servers is a key consideration for data center operators, and understanding servers' power consumption under different operating conditions is an important aspect of it. In this paper, we present a measurement-based case study of high-throughput computing. We analyze power usage information of an operational data center, combined with focused measurements of power reduction techniques for a representative high-throughput workload. The study points out the obstacles encountered by data center operators in their efforts to minimize energy consumption and carbon emissions, and discusses the impact of server configuration adjustments on the energy consumption of processing jobs. We offer actionable recommendations for decreasing the energy usage of servers, while considering both performance and carbon emissions.

cs.NI↗

arXiv2026

Risk-Based Dynamic Thermal Rating in Distribution Transformers via Probabilistic Forecasting

Low voltage (LV) distribution transformers face accelerating demand growth while replacement lead times and costs continue to rise, making improved utilisation of existing assets essential. Static and conservative protection devices (PDs) in distribution transformers are inflexible and limit the available headroom of the transformer. This paper presents a probabilistic framework for dynamically forecasting optimal thermal protection settings. The proposed approach directly predicts the day-ahead scale factor which maximises the dynamic thermal rating of the transformer from historical load, temperature, and metadata using clustered quantile regression models trained on 644 UK LV transformers. Probabilistic forecasting quantifies overheating risk directly through the prediction percentile, enabling risk-informed operational decisions. Results show a 10--12\% additional capacity gain compared to static settings, with hotspot temperature risk matching the selected percentile, including under realistic temperature forecast errors. These results demonstrate a practical approach for distribution network operators to take advantage of PDs with adaptive settings to maximise capacity and manage risk on operational time scales.

eess.SY↗

arXiv2022

SUB-Depth: Self-distillation and Uncertainty Boosting Self-supervised Monocular Depth Estimation

We propose SUB-Depth, a universal multi-task training framework for self-supervised monocular depth estimation (SDE). Depth models trained with SUB-Depth outperform the same models trained in a standard single-task SDE framework. By introducing an additional self-distillation task into a standard SDE training framework, SUB-Depth trains a depth network, not only to predict the depth map for an image reconstruction task, but also to distill knowledge from a trained teacher network with unlabelled data. To take advantage of this multi-task setting, we propose homoscedastic uncertainty formulations for each task to penalize areas likely to be affected by teacher network noise, or violate SDE assumptions. We present extensive evaluations on KITTI to demonstrate the improvements achieved by training a range of existing networks using the proposed framework, and we achieve state-of-the-art performance on this task. Additionally, SUB-Depth enables models to estimate uncertainty on depth output.

cs.CV↗

arXiv2022

2T-POT Hawkes model for left- and right-tail conditional quantile forecasts of financial log-returns: out-of-sample comparison of conditional EVT models

Conditional extreme value theory (EVT) methods promise enhanced forecasting of the extreme tail events that often dominate systemic risk. We present an improved two-tailed peaks-over-threshold (2T-POT) Hawkes model that is adapted for conditional quantile forecasting in both the left and right tails of a univariate time series. This is applied to the daily log-returns of six large cap indices. We also take the unique step of fitting the model at multiple exceedance thresholds (from the 1.25% to 25.00% mirrored quantiles). Quantitatively similar asymmetries in Hawkes parameters are found across all six indices, adding further empirical support to a temporal leverage effect in financial price time series in which the impact of losses is not only larger but also more immediate. Out-of-sample backtests find that our 2T-POT Hawkes model is more reliably accurate than the GARCH-EVT model when forecasting (mirrored) value-at-risk and expected shortfall at the 5% coverage level and below. This suggests that asymmetric Hawkes-type arrival dynamics are a better approximation of the true data generating process for extreme daily log-returns than GARCH-type conditional volatility; our 2T-POT Hawkes model therefore presents a better performing alternative for financial risk modelling.

q-fin.ST↗

arXiv2022

Comparing Generator Unavailability Models with Empirical Distributions from Open Energy Datasets

The modelling of power station outages is an integral part of power system planning. In this work, models of the unavailability of the fleets of eight countries in Northwest Europe are constructed and subsequently compared against empirical distributions derived using data from the open-access ENTSO-e Transparency Platform. Summary statistics of non-sequential models highlight limitations with the empirical modelling, with very variable results across countries. Additionally, analysis of time sequential models suggests a clear need for fleet-specific analytic model parameters. Despite a number of challenges and ambiguities associated with the empirical distributions, it is suggested that a range of valuable qualitative and quantitative insights can be gained by comparing these two complementary approaches for modelling and understanding generator unavailabilities.

eess.SY↗

arXiv2022

Optimization-based modelling and game-theoretic framework for techno-economic analysis of demand-side flexibility: a real case study

This paper proposes a two-step framework for techno-economic analysis of a demand-side flexibility service in distribution networks. Step one applies optimization-based modelling to propose a generic problem formulation which determines the offer curve, in terms of available flexible capacity and its marginal cost, for flexible distribution-connected assets. These offer curves form an input to the second step, which uses a multi-agent iterative game framework to determine the benefits of demand-side flexibility for the Distribution System Operator (DSO) and the service providers. The combined two-step framework simultaneously accounts for the objectives of each flexibility provider, technical constraints of flexible assets, customer preferences, market clearing mechanisms, and strategic bidding by service providers, omission of any of which can lead to erroneous results. The proposed two-step framework has been applied to a real case study in the North East of England to examine four market mechanisms and three bidding strategies. The results showed that among all considered market mechanisms, flexibility markets that operate under discriminatory pricing, such as pay-as-bid and Dutch reverse auctions, are prone to manipulations, especially in the lack of competition. In contrast, uniform pricing pay-as-cleared auction provides limited opportunities for manipulation even when competition is low.

eess.SY↗

arXiv2021

Self-Supervised Monocular Depth Estimation with Internal Feature Fusion

Self-supervised learning for depth estimation uses geometry in image sequences for supervision and shows promising results. Like many computer vision tasks, depth network performance is determined by the capability to learn accurate spatial and semantic representations from images. Therefore, it is natural to exploit semantic segmentation networks for depth estimation. In this work, based on a well-developed semantic segmentation network HRNet, we propose a novel depth estimation network DIFFNet, which can make use of semantic information in down and upsampling procedures. By applying feature fusion and an attention mechanism, our proposed method outperforms the state-of-the-art monocular depth estimation methods on the KITTI benchmark. Our method also demonstrates greater potential on higher resolution training data. We propose an additional extended evaluation strategy by establishing a test set of challenging cases, empirically derived from the standard benchmark.

cs.CV↗

arXiv2021

Asymmetric excitation of left- and right-tail extreme events probed using a Hawkes model: application to financial returns

We construct a two-tailed peak-over-threshold Hawkes model that captures asymmetric self- and cross-excitation in and between left- and right-tail extreme values within a time series. We demonstrate its applicability by investigating extreme gains and losses within the daily log-returns of the S&P 500 equity index. We find that the arrivals of extreme losses and gains are described by a common conditional intensity to which losses contribute twice as much as gains. However, the contribution of the former decays almost five times more quickly than that of the latter. We attribute these asymmetries to the different reactions of market traders to extreme upward and downward movements of asset prices: an example of negativity bias, wherein trauma is more salient than euphoria.

q-fin.ST↗

arXiv2021

Impacts of Heat Decarbonisation on System Adequacy considering Increased Meteorological Sensitivity

This paper explores the impacts of decarbonisation of heat on demand and subsequently on the generation capacity required to secure against system adequacy standards. Gas demand is explored as a proxy variable for modelling the electrification of heating demand in existing housing stock, with a focus on impacts on timescales of capacity markets (up to four years ahead). The work considers the systemic changes that electrification of heating could introduce, including biases that could be introduced if legacy modelling approaches continue to prevail. Covariates from gas and electrical regression models are combined to form a novel, time-collapsed system model, with demand-weather sensitivities determined using lasso-regularized linear regression. It is shown, using a GB case study with one million domestic heat pump installations per year, that the sensitivity of electrical system demand to temperature (and subsequently sensitivities to cold/warm winter seasons) could increase by 50% following four years of heat demand electrification. A central estimate of 1.75 kW additional peak demand per heat pump is estimated, with variability across three published heat demand profiles leading to a range of more than 14 GW in the most extreme cases. It is shown that the legacy approach of scaling historic demand, as compared to the explicit modelling of heat, could lead to over-procurement of 0.79 GW due to bias in estimates of additional capacity to secure. Failure to address this issue could lead to £100m overspend on capacity over ten years.

stat.AP↗

arXiv2020

Hybrid European MV-LV Network Models for Smart Distribution Network Modelling

A pair of European-style, integrated MV-LV circuits are presented, created by combining generic MV and real LV networks. The two models have 86,000 and 113,000 nodes, and are made readily available for download in the OpenDSS file format. Primary substation tap change controls and MV-LV feeders are represented as three-phase unbalanced distribution network models, capturing the coupling of voltages at the MV level. The assumptions made in constructing the models are outlined, including a preconditioning step that reduces the number of nodes by more than five times without affecting the solution. Two flexibility-based case studies are presented, with TSO-DSO and peer-peer-based smart controls considered. The demonstration of the heterogeneous nature of these systems is corroborated by the analysis of measured LV voltage data. The models are intended to aid the development of algorithms for maximising the benefits of smart devices within the context of whole energy systems.

eess.SY↗

arXiv2011

Responsibility Modeling for the Sociotechnical Risk Analysis of Coalitions of Systems

Society is challenging systems engineers by demanding ever more complex and integrated systems. With the rise of cloud computing and systems-of-systems (including cyber-physical systems) we are entering an era where mission critical services and applications will be dependent upon 'coalitions-of-systems'. Coalitions-of-systems (CoS) are a class of system similar to systems-of-systems but they differ in that they interact to further overlapping self-interests rather than an overarching mission. Assessing the sociotechnical risks associated with CoS is an open research question of societal importance as existing risk analysis techniques typically focus on the technical aspects of systems and ignore risks associated with coalition partners reneging on responsibilities or leaving the coalition. We demonstrate that a responsibility modeling based risk analysis approach enables the identification of sociotechnical risks associated with CoS. The approach identifies hazards and associated risks that may arise when relying upon a coalition of human/organizational/technical agents to provision a service or application. Through a case study of a proposed cloud IT infrastructure migration we show how the technique identifies vulnerabilities that may arise because of human, organizational or technical agents failing to discharge responsibilities.

cs.SE↗

arXiv2011

Expectations and Reality: Why an enterprise software system didn't work as planned

Over two decades, we and other research groups have found that ethnographic and social analyses of work settings can provide insights useful to the process of system analysis and design. Despite this, ethnographic and social analyses have not been widely assimilated into industry practice. Practitioners tend to address sociotechnical factors in an ad-hoc manner, often post-implementation, once system use or outcome has become problematic. In response to this, we have developed a lightweight qualitative approach to provide insights to ameliorate problematic system deployments. Unlike typical ethnographies and social analyses of work activity that inform systems analysis and design; we argue that analysis of intentional and structural factors to inform system deployment and integration can have a shorter time duration and yet can provide actionable insights. We evaluate our approach using a case study of a problematic enterprise document manage-ment system within a multinational systems engineering organization. Our find-ings are of academic and practical significance as our approach demonstrates that structural-intentional analysis scales to enable the timely analysis of large-scale system deployments.

cs.SE↗

arXiv2010

The Cloud Adoption Toolkit: Supporting Cloud Adoption Decisions in the Enterprise

Cloud computing promises a radical shift in the provisioning of computing resource within the enterprise. This paper describes the challenges that decision makers face when assessing the feasibility of the adoption of cloud computing in their organisations, and describes our Cloud Adoption Toolkit, which has been developed to support this process. The toolkit provides a framework to support decision makers in identifying their concerns, and matching these concerns to appropriate tools/techniques that can be used to address them. Cost Modeling is the most mature tool in the toolkit, and this paper shows its effectiveness by demonstrating how practitioners can use it to examine the costs of deploying their IT systems on the cloud. The Cost Modeling tool is evaluated using a case study of an organization that is considering the migration of some of its IT systems to the cloud. The case study shows that running systems on the cloud using a traditional "always on" approach can be less cost effective, and the elastic nature of the cloud has to be used to reduce costs. Therefore, decision makers have to be able to model the variations in resource usage and their systems deployment options to obtain accurate cost estimates.

cs.DC↗

arXiv2010

Lessons from the Failure and Subsequent Success of a Complex Healthcare Sector IT Project

This paper argues that IT failures diagnosed as errors at the technical or project management level are often mistakenly pointing to symptoms of failure rather than a project's underlying socio-complexity (complexity resulting from the interactions of people and groups) which is usually the actual source of failure. We propose a novel method, Stakeholder Impact Analysis, that can be used to identify risks associated with socio-complexity as it is grounded in insights from the social sciences, psychology and management science. This paper demonstrates the effectiveness of Stakeholder Impact Analysis by using the 1992 London Ambulance Service Computer Aided Dispatch project as a case study, and shows that had our method been used to identify the risks and had they been mitigated, it would have reduced the risk of project failure. This paper's original contribution comprises expanding upon existing accounts of failure by examining failures at a level of granularity not seen elsewhere that enables the underlying socio-complexity sources of risk to be identified.

cs.SE↗

arXiv2010

The Cloud Adoption Toolkit: Addressing the Challenges of Cloud Adoption in Enterprise

Cloud computing promises a radical shift in the provisioning of computing resource within the enterprise. This paper: i) describes the challenges that decision makers face when attempting to determine the feasibility of the adoption of cloud computing in their organisations; ii) illustrates a lack of existing work to address the feasibility challenges of cloud adoption in the enterprise; iii) introduces the Cloud Adoption Toolkit that provides a framework to support decision makers in identifying their concerns, and matching these concerns to appropriate tools/techniques that can be used to address them. The paper adopts a position paper methodology such that case study evidence is provided, where available, to support claims. We conclude that the Cloud Adoption Toolkit, whilst still under development, shows signs that it is a useful tool for decision makers as it helps address the feasibility challenges of cloud adoption in the enterprise.

cs.DC↗

arXiv2010

Cloud Migration: A Case Study of Migrating an Enterprise IT System to IaaS

This case study illustrates the potential benefits and risks associated with the migration of an IT system in the oil & gas industry from an in-house data center to Amazon EC2 from a broad variety of stakeholder perspectives across the enterprise, thus transcending the typical, yet narrow, financial and technical analysis offered by providers. Our results show that the system infrastructure in the case study would have cost 37% less over 5 years on EC2, and using cloud computing could have potentially eliminated 21% of the support calls for this system. These findings seem significant enough to call for a migration of the system to the cloud but our stakeholder impact analysis revealed that there are significant risks associated with this. Whilst the benefits of using the cloud are attractive, we argue that it is important that enterprise decision-makers consider the overall organizational implications of the changes brought about with cloud computing to avoid implementing local optimizations at the cost of organization-wide performance.

cs.DC↗