A5100456490- Caching and Content Delivery
- Opportunistic and Delay-Tolerant Networks
- Peer-to-Peer Network Technologies
- Privacy-Preserving Technologies in Data
- Cooperative Communication and Network Coding
- IoT and Edge/Fog Computing
- Software-Defined Networks and 5G
- Cloud Computing and Resource Management
- Neural Networks and Applications
- Internet Traffic Analysis and Secure E-voting
- Advanced Data Storage Technologies
- Green IT and Sustainability
- Natural Language Processing Techniques
- Adversarial Robustness in Machine Learning
- Network Security and Intrusion Detection
- Covalent Organic Framework Applications
- Age of Information Optimization
- Spam and Phishing Detection
- Software System Performance and Reliability
- Advanced Neural Network Applications
- Topic Modeling
- Vehicular Ad Hoc Networks (VANETs)
- Big Data and Business Intelligence
- Recommender Systems and Techniques
- Sharing Economy and Platforms
Microsoft Research (United Kingdom)
2023
Microsoft (United States)
2023
University of Cambridge
2013-2022
Northwestern Polytechnical University
2022
China Geological Survey
2022
Shaanxi Normal University
2022
Beijing Jiaotong University
2020
Shanghai Electric (China)
2020
Xuzhou Medical College
2018
Shanxi Normal University
2016
The incorporation of dynamic voltage scaling technology into computation offloading offers more flexibilities for mobile edge computing. In this paper, we investigate partial by jointly optimizing the computational speed smart device (SMD), transmit power SMD, and ratio with two system design objectives: energy consumption SMD minimization (ECM) latency application execution (LM). Considering case that is served a single cloud server, formulate both ECM problem LM as nonconvex problems. To...
This paper introduces a simple yet effective query expansion approach, denoted as query2doc, to improve both sparse and dense retrieval systems. The proposed method first generates pseudo-documents by few-shot prompting large language models (LLMs), then expands the with generated pseudo documents. LLMs are trained on web-scale text corpora adept at knowledge memorization. from often contain highly relevant information that can aid in disambiguation guide retrievers. Experimental results...
Recent research has shown a substantial active presence of bots in online social networks (OSNs). In this paper we utilise our previous work (Stweeler) to comparatively analyse the usage and impact humans on Twitter, one largest OSNs world. We collect large-scale Twitter dataset define various metrics based tweet metadata. Using human annotation task assign 'bot' 'human' ground-truth labels dataset, compare annotations against an bot detection tool for evaluation. then ask series questions...
This paper outlines the IoT Databox model as a means of making Internet Things (IoT) accountable to individuals. Accountability is key building consumer trust and mandated by European Union's general data protection regulation (GDPR). We focus here on 'external' subject accountability requirement specified GDPR how meeting this turns surfacing invisible actions interactions connected devices social arrangements in which they are embedded. The proposed an principle enabling providing...
Peer-to-peer networks have been quite thoroughly measured over the past years, however it is interesting to note that BitTorrent Mainline DHT has received very little attention even though by far largest of currently active overlay systems, as our results show. As differs from other existing measurement methodologies are not appropriate for studying it. In this paper we present an efficient methodology estimating number users in network. We identified omission previous used measure size...
We are all increasingly the subjects of data collection and processing systems that use generated both about by us to provide optimise a wide range services. Means for others collect process concerns each -- often referred possessively as "your data" only increasing with long-heralded advent Internet Things just latest example. As result, means enable personal management is generally recognised pressing societal issue.
Scoped-flooding is a technique for content discovery in broad networking context. This paper investigates the effects of scoped-flooding on various topologies information- centric networking. Using proposed ring model, we show that flooding can be constrained within very small neighbourhood to achieve most gains which come from areas where growth rate relatively low, i.e., net- work edge. We also study two strategies and com- pare their behaviours. Given caching schemes favour more popular...
The WWW has seen a massive growth in variety and usage of OSNs. rising population users on Twitter its open nature made it an ideal platform for various kinds opportunistic pursuits, such as news emergency communication, business promotion, political campaigning, spamming spreading malicious content. Most these pursuits are exploited through automated programs, known bots. In this study we propose framework (Stweeler) to bot impact influence from systems social media perspectives.
Emerging Machine Learning (ML) techniques, such as Deep Neural Network, are widely used in today's applications and services. However, with social awareness of privacy personal data rapidly rising, it becomes a pressing challenging societal issue to both keep private benefit from the analytics power ML techniques at same time. In this paper, we argue that avoid those costs, reduce latency processing, minimise raw revealed service providers, many future AI services could be deployed on users'...
Instead of simply matching a query to pre-existing passages, generative retrieval generates identifier strings passages as the target. At cost, must be distinctive enough represent passage. Current approaches use either numeric ID or text piece (such title substrings) identifier. However, these identifiers cannot cover passage's content well. As such, we are motivated propose new type identifier, synthetic identifiers, that generated based on passage and could integrate contextualized...
Distributed hash tables (DHT) are a key building block for modern P2P content-distribution system, example in implementing the distributed tracker of BitTorrent Mainline DHT. DHTs, due to their fully nature, known be vulnerable certain kinds attacks and different defenses have been proposed against these attacks. In this paper, we consider two on DHT, one already attack new kind an attack, show how they can targeted We complement them by extensive measurement study using honeypots which...
Do-It-Yourself (DIY) networks are decentralised built by an (often) amateur community. As DIY do not rely on the need for backhaul Internet connectivity, these mostly a mix of both offline and online networks. Although have their own homegrown services, current Internet-based cloud services often useful, access to some could be beneficial Considering that most challenged migrating service virtualisation instances face great challenges. Service Centric Networking (SCN) has been recently...
Information-centric networking (ICN) is a popular research topic. At its heart the concept of in-network caching. Various algorithms have been proposed for optimizing ICN caching, many which rely on collaborative principles, i.e. multiple caches interacting to decide what store. Past work has assumed altruistic nodes that will sacrifice their own performance global optimum. We argue this assumption insufficient and oversimplifies reality. address problem by modeling caching as Nash...
The Internet is overloading its users with excessive information flows, so that effective content-based filtering becomes crucial in improving user experience and work efficiency. Latent semantic analysis has long been demonstrated as a promising retrieval technique to search for relevant articles from large text corpora. We build Kvasir, recommendation system, on top of latent other state-of-the-art technologies seamlessly integrate an automated proactive content provision service into web...
Geographic detectors can quickly detect spatial stratified heterogeneity and quantitatively reveal the intensity of driving factors heavy metal content, which is great significance for prevention, control, remediation soil pollution. In order to differentiation influencing content on town-scale, 788 topsoil samples were collected from a town in hinterland Chengdu Plain. Soil (Cd, Hg, As, Cu, Pb, Cr, Zn, Ni) pollution risk assessments carried out by using geo-accumulation index method....
Community networks owned and operated by local communities have recently gained popularity as a low cost solution for Internet access. In this paper, we seek to understand the characteristics of usage in community provide useful insights on designing improving rural areas. We report results socio-technical study carried out during three month measurement wireless mesh network (CWMN) which has been operating two years area northern Thailand. An on-site social interview was also conducted...
We describe an agent-based framework that successfully simulates the key aspects of most famous flash crash in history: Flash Crash May 6, 2010. In our model, market making high-frequency traders collectively create a feedback loop system triggered by large institutional sell, consistent with widely cited "hot-potato effect." With help simulations, we discover functional relationships between number HFT makers, their inventory sizes or speed, and probability another similar crash. The model...
Caching is a core principle of information-centric networking (ICN). Many novel algorithms have been proposed for enabling ICN caching, many which rely on collaborative principles, i.e. multiple caches interacting to decide what store. Past work has assumed entirely altruistic nodes that will sacrifice their own performance the global optimum. In this paper, we argue assumption flawed. We address problem by modelling in-network caching as Nash bargaining game. develop optimal and heuristic...
Many current Internet services rely on inferences from models trained user data. Commonly, both the training and inference tasks are carried out using cloud resources fed by personal data collected at scale users. Holding such large collections of in creates privacy risks to subjects, but is currently required for users benefit services. We explore how provide model a system where computation pushed preference moving cloud, obviating many risks. Specifically, we take an initial learnt small...
Information-centric networks are a new paradigm for addressing and accessing content on the Internet, with Content-Centric Networking (CCN) being one of more popular candidate solutions. CCN de-couples from location it is hosted allows mobility node requesting content. However, CCN's ability to handle source limited so far little research has focused how both endpoints would be able mobile. We focus source, using network embeddings as tool. Network have already been proposed management in...
Scoped-flooding is used for content discovery in a broad networking context and it has significant impact on the design of caching algorithms communication network. Despite its wide usage, thorough analysis how scoped-flooding affects network's performance, e.g., efficiency, missing. To develop better understanding, we first model behavior by help theoretical network growth utility. Next, investigate effects various topologies information-centric networks (ICNs). Using proposed ring model,...