Search

Software services are increasingly migrating to the cloud, requiring trust in actors with direct access to the hardware, software and data comprising the service. A distributed datastore storing critical data sits at the core of many services; a prime example being etcd in Kubernetes. Trusted execution environments can secure this data from cloud providers during execution, but it is complex to build trustworthy data storage systems using such mechanisms. We present the design and evaluation of the Ledger-backed Secure Key-Value datastore (LSKV), a distributed datastore that provides an etcd-like API but can use trusted execution mechanisms to keep cloud providers outside the trust boundary. LSKV provides a path to transition traditional systems towards confidential execution, provides competitive performance compared to etcd, and helps clients to gain trust in intermediary services. LSKV forms a foundational core, lowering the barriers to building more trustworthy systems.

Application tail latency is a key metric for many services, with high latencies being linked directly to loss of revenue. Modern deeply-nested micro-service architectures exacerbate tail latencies, increasing the likelihood of users experiencing them. In this work, we show how CPU overcommitment by OS threads leads to high tail latencies when applications are under heavy load. CPU overcommitment can arise from two operational factors: incorrectly determining the number of CPUs available when under a CPU quota, and the ignorance of neighbour applications and their CPU usage. We discuss different languages' solutions to obtaining the CPUs available, evaluating the impact, and discuss opportunities for a more unified language-independent interface to obtain the number of CPUs available. We then evaluate the impact of neighbour usage on tail latency and introduce a new neighbour-aware threadpool, the friendlypool, that dynamically avoids overcommitment. In our evaluation, the friendlypool reduces maximum worker latency by up to $6.7\times$ at the cost of decreasing throughput by up to $1.4\times$.

The rapid adoption of large language models (LLMs) has led to significant advances in natural language processing and text generation. However, the energy consumed through LLM model inference remains a major challenge for sustainable AI deployment. To address this problem, we model the workload-dependent energy consumption and runtime of LLM inference tasks on heterogeneous GPU-CPU systems. By conducting an extensive characterization study of several state-of-the-art LLMs and analyzing their energy and runtime behavior across different magnitudes of input prompts and output text, we develop accurate (R^2>0.96) energy and runtime models for each LLM. We employ these models to explore an offline, energy-optimal LLM workload scheduling framework. Through a case study, we demonstrate the advantages of energy and accuracy aware scheduling compared to existing best practices., Comment: 7 pages, appearing at HotCarbon 2024

Both the training and use of Large Language Models (LLMs) require large amounts of energy. Their increasing popularity, therefore, raises critical concerns regarding the energy efficiency and sustainability of data centers that host them. This paper addresses the challenge of reducing energy consumption in data centers running LLMs. We propose a hybrid data center model that uses a cost-based scheduling framework to dynamically allocate LLM tasks across hardware accelerators that differ in their energy efficiencies and computational capabilities. Specifically, our workload-aware strategy determines whether tasks are processed on energy-efficient processors or high-performance GPUs based on the number of input and output tokens in a query. Our analysis of a representative LLM dataset, finds that this hybrid strategy can reduce CPU+GPU energy consumption by 7.5% compared to a workload-unaware baseline.

Global financial crime activity is driving demand for machine learning solutions in fraud prevention. However, prevention systems are commonly serviced to financial institutions in isolation, and few provisions exist for data sharing due to fears of unintentional leaks and adversarial attacks. Collaborative learning advances in finance are rare, and it is hard to find real-world insights derived from privacy-preserving data processing systems. In this paper, we present a collaborative deep learning framework for fraud prevention, designed from a privacy standpoint, and awarded at the recent PETs Prize Challenges. We leverage latent embedded representations of varied-length transaction sequences, along with local differential privacy, in order to construct a data release mechanism which can securely inform externally hosted fraud and anomaly detection models. We assess our contribution on two distributed data sets donated by large payment networks, and demonstrate robustness to popular inference-time attacks, along with utility-privacy trade-offs analogous to published work in alternative application domains.

With the promise of federated learning (FL) to allow for geographically-distributed and highly personalized services, the efficient exchange of model updates between clients and servers becomes crucial. FL, though decentralized, often faces communication bottlenecks, especially in resource-constrained scenarios. Existing data compression techniques like gradient sparsification, quantization, and pruning offer some solutions, but may compromise model performance or necessitate expensive retraining. In this paper, we introduce FedSZ, a specialized lossy-compression algorithm designed to minimize the size of client model updates in FL. FedSZ incorporates a comprehensive compression pipeline featuring data partitioning, lossy and lossless compression of model parameters and metadata, and serialization. We evaluate FedSZ using a suite of error-bounded lossy compressors, ultimately finding SZ2 to be the most effective across various model architectures and datasets including AlexNet, MobileNetV2, ResNet50, CIFAR-10, Caltech101, and Fashion-MNIST. Our study reveals that a relative error bound 1E-2 achieves an optimal tradeoff, compressing model states between 5.55-12.61x while maintaining inference accuracy within <0.5% of uncompressed results. Additionally, the runtime overhead of FedSZ is <4.7% or between of the wall-clock communication-round time, a worthwhile trade-off for reducing network transfer times by an order of magnitude for networks bandwidths <500Mbps. Intriguingly, we also find that the error introduced by FedSZ could potentially serve as a source of differentially private noise, opening up new avenues for privacy-preserving FL., Comment: Appearing at 44th IEEE International Conference on Distributed Computing Systems (ICDCS)

In the edge environment servers are no longer being co-located away from clients, instead they are being co-located with clients away from other servers, focusing on reliable and performant operation. Orchestration platforms, such as Kubernetes, are a key system being transitioned to the edge but they remain unsuited to the environment, stemming primarily from their critical key-value stores. In this work we derive requirements from the edge environment showing that, fundamentally, the design of distributed key-value datastores, such as etcd, is unsuited to meet them. Using these requirements, we explore the design space for distributed key-value datastores and implement two successive mutations of etcd for different points: mergeable-etcd and dismerge, trading linearizability for causal consistency based on CRDTs. mergeable-etcd retains the linear revision history but encounters inherent shortcomings, whilst dismerge embraces the causal model. Both stores are local-first, maintaining reliable performance under network partitions and variability, drastically surpassing etcd's performance, whilst maintaining competitive performance in reliable settings.

As the prevalence of Internet-of-Things (IoT) devices becomes more and more dominant, so too do the associated management and security challenges. One such challenge is the exploitation of vulnerable devices for recruitment into botnets, which can be used to carry out Distributed Denial-of-Service (DDoS) attacks. The recent Manufacturer Usage Description (MUD) standard has been proposed as a way to mitigate this problem, by allowing manufacturers to define communication patterns that are permitted for their IoT devices, with enforcement at the gateway home router. In this paper, we present a novel integrated system implementation that uses a MUD manager (osMUD) to parse an extended set of MUD rules, which also allow for rate-limiting of traffic and for setting appropriate thresholds. Additionally, we present two new backends for MUD rule enforcement, one based on eBPF and the other based on the Linux standard iptables. The reported evaluation results show that these techniques are feasible and effective in protecting against attacks and in terms of their impact on legitimate traffic and on the home gateway., Comment: 12 pages, 10 figures, in review currently

The report describes the digital architecture developed for the West Cambridge Digital Twin, particularly focussed on real-time sensor data collection and analysis with a privacy framework allowing occupants of the buildings to be first-class participants in the system. The implementation has some notable characteristics. In particular 'push' technology is used throughout such that information streams from the incoming asynchronous individual sensor events through to the end-user web pages with the absolute minimum latency, including real-time generated simple and complex events derived from the the underlying sensor data and the updating of visualisations such as an in-building heatmap. We believe the ability of the entire system to respond in the timescale of individual sensor messages to be unique. JSON structures are used to represent all data types including sensor readings, sensor types, building objects, organisations and people, with the idea that JSON-LD may represent a more suitable way than XML/RDF for managing relations between those objects (such as the 'occupies' relationship of people to offices, or the 'type' relationship of sensors to sensor types)., Comment: 72 pages. arXiv admin note: substantial text overlap with arXiv:2103.04924, arXiv:2103.09169

Given a simple concept that has the potential for a methodological contribution to the field of HCI, the constraints of submission conventions within the field may make it impossible to communicate the concept in a manner that is intelligible to the reader., Comment: 9 pages

This chapter draws from across the foregoing chapters discussing many core HDI approaches and disciplinary perspectives to consider the specific application of HDI in home network security. While much work has considered the challenges of securing in home IoT devices and their communications, especially for those with limited power or computational capacity, scant attention has been paid by the research community to home network security, and its acceptability and usability, from the viewpoint of ordinary citizens. It will be clear that we need a radical transformation in our approach to designing domestic networking infrastructure to guard against widespread cyber-attacks that threaten to counter the benefits of the IoT. Our aim has to be to defend against enemies inside the walls, to protect critical functionality in the home against rogue devices and prevent the proliferation of disruptive wide-scale IoT DDOS attacks that are already occurring [1]., Comment: Preprint of Chapter 9 of Privacy by Design for the Internet of Things: Building accountability and security

Machine Learning (ML) techniques have begun to dominate data analytics applications and services. Recommendation systems are a key component of online service providers. The financial industry has adopted ML to harness large volumes of data in areas such as fraud detection, risk-management, and compliance. Deep Learning is the technology behind voice-based personal assistants, etc. Deployment of ML technologies onto cloud computing infrastructures has benefited numerous aspects of our daily life. The advertising and associated online industries in particular have fuelled a rapid rise the in deployment of personal data collection and analytics tools. Traditionally, behavioural analytics relies on collecting vast amounts of data in centralised cloud infrastructure before using it to train machine learning models that allow user behaviour and preferences to be inferred. A contrasting approach, distributed data analytics, where code and models for training and inference are distributed to the places where data is collected, has been boosted by two recent, ongoing developments: increased processing power and memory capacity available in user devices at the edge of the network, such as smartphones and home assistants; and increased sensitivity to the highly intrusive nature of many of these devices and services and the attendant demands for improved privacy. Indeed, the potential for increased privacy is not the only benefit of distributing data analytics to the edges of the network: reducing the movement of large volumes of data can also improve energy efficiency, helping to ameliorate the ever increasing carbon footprint of our digital infrastructure, enabling much lower latency for service interactions than is possible when services are cloud-hosted. These approaches often introduce challenges in privacy, utility, and efficiency trade-offs, while having to ensure fruitful user engagement., Comment: Accepted as Chapter 8 of "Privacy by Design for the Internet of Things: Building accountability and security"

Distributed consensus, the ability to reach agreement in the face of failures, is a fundamental primitive for constructing reliable distributed systems. The Paxos algorithm is synonymous with consensus and widely utilized in production. Paxos uses two phases: phase one and phase two, each requiring a quorum of acceptors, to reach consensus during a round of the protocol. Traditionally, Paxos requires that all quorums, regardless of phase or round, intersect and majorities are often used for this purpose. Flexible Paxos proved that it is only necessary for phase one quorum of a given round to intersect with the phase two quorums of all previous rounds. In this paper, we re-examine how Paxos approaches the problem of consensus. We look again at quorum intersection in Flexible Paxos and observe that quorum intersection can be safely weakened further. Most notably, we observe that if a proposer learns that a value was proposed in some previous round then its phase one no longer needs to intersect with the phase two quorums from that round or from any previous rounds. Furthermore, in order to provide an intuitive explanation of our results, we propose a novel abstraction for reasoning about Paxos which utilizes write-once registers., Comment: to be published in the 9th Workshop on Principles and Practice of Consistency for Distributed Data (PaPoC'22)

Internet-of-Things (IoT) devices are known to be the source of many security problems, and as such, they would greatly benefit from automated management. This requires robustly identifying devices so that appropriate network security policies can be applied. We address this challenge by exploring how to accurately identify IoT devices based on their network behavior, while leveraging approaches previously proposed by other researchers. We compare the accuracy of four different previously proposed machine learning models (tree-based and neural network-based) for identifying IoT devices. We use packet trace data collected over a period of six months from a large IoT test-bed. We show that, while all models achieve high accuracy when evaluated on the same dataset as they were trained on, their accuracy degrades over time, when evaluated on data collected outside the training set. We show that on average the models' accuracy degrades after a couple of weeks by up to 40 percentage points (on average between 12 and 21 percentage points). We argue that, in order to keep the models' accuracy at a high level, these need to be continuously updated., Comment: To appear in TMA 2021 conference. 9 pages, 6 figures. arXiv admin note: text overlap with arXiv:2011.08605

Recent years have seen Kubernetes emerge as a primary choice for container orchestration. Kubernetes largely targets the cloud environment but new use cases require performant, available and scalable orchestration at the edge. Kubernetes stores all cluster state in etcd, a strongly consistent key-value store. We find that at larger etcd cluster sizes, offering higher availability, write request latency significantly increases and throughput decreases similarly. Coupled with approximately 30% of Kubernetes requests being writes, this directly impacts the request latency and availability of Kubernetes, reducing its suitability for the edge. We revisit the requirement of strong consistency and propose an eventually consistent approach instead. This enables higher performance, availability and scalability whilst still supporting the broad needs of Kubernetes. This aims to make Kubernetes much more suitable for performance-critical, dynamically-scaled edge solutions., Comment: 6 pages. Accepted in EdgeSys '21

Building Management Systems (BMSs) have evolved in recent years, in ways that require changes to existing network architectures that follow the store-then-analyse approach. The primary cause is the increasing deployment of a diverse range of cost-effective sensors and actuators in smart buildings that generate real-time streaming data. Any in-building system with a large number of sensors needs a framework for real-time data collection and concurrent stream processing from sensors connected using a range of networks. We present SenseRT, a system for managing and analysing in-building real-time streams of sensor data. SenseRT collects streams of real-time data from sensors connected using a range of network protocols. It supports concurrent modules simultaneously performing stream processing over real-time data, asynchronously and non-blocking, with results made available with minimal latency. We describe a prototype implementation deployed in two University department buildings, demonstrating its effectiveness.

Legacy Building Information Modelling (BIM) systems are not designed to process the high-volume, high-velocity data emitted by in-building Internet-of-Things (IoT) sensors. Historical lack of consideration for the real-time nature of such data means that outputs from such BIM systems typically lack the timeliness necessary for enacting decisions as a result of patterns emerging in the sensor data. Similarly, as sensors are increasingly deployed in buildings, antiquated Building Management Systems (BMSs) struggle to maintain functionality as interoperability challenges increase. In combination these motivate the need to fill an important gap in smart buildings research, to enable faster adoption of these technologies, by combining BIM, BMS and sensor data. This paper describes the data architecture of the Adaptive City Platform, designed to address these combined requirements by enabling integrated BIM and real-time sensor data analysis across both time and space.

Internet-of-Things (IoT) devices are known to be the source of many security problems, and as such they would greatly benefit from automated management. This requires robustly identifying devices so that appropriate network security policies can be applied. We address this challenge by exploring how to accurately identify IoT devices based on their network behavior, using resources available at the edge of the network. In this paper, we compare the accuracy of five different machine learning models (tree-based and neural network-based) for identifying IoT devices by using packet trace data from a large IoT test-bed, showing that all models need to be updated over time to avoid significant degradation in accuracy. In order to effectively update the models, we find that it is necessary to use data gathered from the deployment environment, e.g., the household. We therefore evaluate our approach using hardware resources and data sources representative of those that would be available at the edge of the network, such as in an IoT deployment. We show that updating neural network-based models at the edge is feasible, as they require low computational and memory resources and their structure is amenable to being updated. Our results show that it is possible to achieve device identification and categorization with over 80% and 90% accuracy respectively at the edge., Comment: 13 pages, 8 figures, 4 tables

Paxos, the de facto standard approach to solving distributed consensus, operates in two phases, each of which requires an intersecting quorum of nodes. Multi-Paxos reduces this to one phase by electing a leader but this leader is also a performance bottleneck. Fast Paxos bypasses the leader but has stronger quorum intersection requirements. In this paper we observe that Fast Paxos' intersection requirements can be safely relaxed, reducing to just one additional intersection requirement between phase-1 quorums and any pair of fast round phase-2 quorums. We thus find that the quorums used with Fast Paxos are larger than necessary, allowing alternative quorum systems to obtain new tradeoffs between performance and fault-tolerance., Comment: To be published in the Proceedings of International Conference on Distributed Computing and Networking 2021 (ICDCN '21)

Due to the advancement of IoT devices in both domestic and industrial environments, the need to incorporate a mechanism to build accountability in the IoT ecosystem is paramount. In the last few years, various initiatives have been started in this direction addressing many socio-technical concerns and challenges to build an accountable system. The solution that has received a lot of attention in both industry and academia is the Manufacturer Usage Description (MUD) specification. It gives the possibility to the IoT device manufacturers to describe communications needed by each device to work properly. MUD implementation is challenging not only due to the diversity of IoT devices and manufacturer/operator/regulators but also due to the incremental integration of MUD-based flow control in the already existing Internet infrastructure. To provide a better understanding of these challenges, in this work, we explore and investigate the prototypes of three implementations proposed by different research teams and organisations, useful for the community to understand which are the various features implemented by the existing technologies. By considering that there exist some behaviours which can be only defined by local policy, we propose a MUD capable network integrating our User Policy Server(UPS). The UPS provides network administrators and endusers an opportunity to interact with MUD components through a user-friendly interface. Hence, we present a comprehensive survey of the challenges., Comment: 7 pages, 5 figures, WIP

Distributed consensus is a fundamental primitive for constructing fault-tolerant, strongly-consistent distributed systems. Though many distributed consensus algorithms have been proposed, just two dominate production systems: Paxos, the traditional, famously subtle, algorithm; and Raft, a more recent algorithm positioned as a more understandable alternative to Paxos. In this paper, we consider the question of which algorithm, Paxos or Raft, is the better solution to distributed consensus? We analyse both to determine exactly how they differ by describing a simplified Paxos algorithm using Raft's terminology and pragmatic abstractions. We find that both Paxos and Raft take a very similar approach to distributed consensus, differing only in their approach to leader election. Most notably, Raft only allows servers with up-to-date logs to become leaders, whereas Paxos allows any server to be leader provided it then updates its log to ensure it is up-to-date. Raft's approach is surprisingly efficient given its simplicity as, unlike Paxos, it does not require log entries to be exchanged during leader election. We surmise that much of the understandability of Raft comes from the paper's clear presentation rather than being fundamental to the underlying algorithm being presented., Comment: To be published in the 7th Workshop on Principles and Practice of Consistency for Distributed Data (PaPoC)

Compiling applications as unikernels allows them to be tailored to diverse execution environments. Dependency on a monolithic operating system is replaced with linkage against libraries that provide specific services. Doing so in practice has revealed a major barrier: managing the configuration matrix across heterogenous execution targets. A realistic unikernel application depends on hundreds of libraries, each of which may place different demands on the different target execution platforms (e.g.,~cryptographic acceleration). We propose a modular approach to structuring large scale codebases that cleanly separates configuration, application and operating system logic. Our implementation is built on the \mirage unikernel framework, using the \ocaml language's powerful abstraction and metaprogramming facilities. Leveraging modules allows us to build many components independently, with only loose coupling through a set of standardised signatures. Components can be parameterized by other components and composed. Our approach accounts for state, dependency ordering, and error management, and our usage over the years has demonstrated significant efficiency benefits by leveraging compiler features such as global link-time optimisation during the configuration process. We describe our application architecture and experiences via some practical applications of our approach, and discuss how library development in \mirage can facilitate adoption in other unikernel frameworks and programming languages.

We continue to see increasingly widespread deployment of IoT devices, with apparent intent to embed them in our built environment likely to accelerate if smart city and related programmes succeed. In this paper we are concerned with the ways in which current generation IoT devices are being designed in terms of their ill-considered dependencies on network connectivity and services. Our hope is to provide evidence that such dependencies need to be better thought through in design, and better documented in implementation so that those responsible for deploying these devices can be properly informed as to the impact of device deployment (at scale) on infrastructure resilience. We believe this will be particularly relevant as we feel that commodity IoT devices are likely to be commonly used to retrofit `smart' capabilities to existing buildings, particularly domestic buildings. To the existing body of work on network-level behaviour of IoT devices, we add \one~a protocol-level breakdown and analysis of periodicity, \two~an exploration of the service and infrastructure dependencies that will implicitly be taken in `smart' environments when IoT devices are deployed, and \three~examination of the robustness of device operation when connectivity is disrupted. We find that many devices make use of services distributed across the planet and thus appear dependent on the global network infrastructure even when carrying out purely local actions. Some devices cease to operate properly without network connectivity (even where their behaviour appears, on the face of it, to require only local information, e.g.,~the Hive thermostat). Further, they exhibit quite different network behaviours, typically involving significantly more traffic and possibly use of otherwise unobserved protocols, when connectivity is recovered after some disruption., Comment: IoTDI '19: Internet of Things Design and Implementation, April 15--18, 2019, Montreal, QC, Canada

To date, cloud applications have used datacenter resources through manual configuration and deployment of virtual machines and containers. Current trends see increasing use of microservices, where larger applications are split into many small containers, to be developed and deployed independently. However, even with the rise of the devops movement and orchestration facilities such as Kubernetes, there is a tendency to separate development from deployment. We present an exploration of a more extreme point on the devops spectrum: Fractal. Developers embed orchestration logic inside their application, fully automating the processes of scaling up and down. Providing a set of extensions to and an API over the Jitsu platform, we outline the design of Fractal and describe the key features of its implementation: how an application is self-replicated, how replica lifecycles are managed, how failure recovery is handled, and how network traffic is transparently distributed between replicas. We present evaluation of a self-scaling website, and demonstrate that Fractal is both useful and feasible.

In this paper, we introduce Zest (REST over ZeroMQ), a middleware technology in support of an Internet of Things (IoT). Our work is influenced by the Constrained Application Protocol (CoAP) but emphasises systems that can support fine-grained access control to both resources and audit information, and can provide features such as asynchronous communication patterns between nodes. We achieve this by using a hybrid approach that combines a RESTful architecture with a variant of a publisher/subscriber topology that has enhanced routing support. The primary motivation for Zest is to provide inter-component communications in the Databox, but it is applicable in other contexts where tight control needs to be maintained over permitted communication patterns.

Distributed consensus, the ability to reach agreement in the face of failures and asynchrony, is a fundamental primitive for constructing reliable distributed systems from unreliable components. The Paxos algorithm is synonymous with distributed consensus, yet it performs poorly in practice and is famously difficult to understand. In this paper, we re-examine the foundations of distributed consensus. We derive an abstract solution to consensus, which utilises immutable state for intuitive reasoning about safety. We prove that our abstract solution generalises over Paxos as well as the Fast Paxos and Flexible Paxos algorithms. The surprising result of this analysis is a substantial weakening to the quorum requirements of these widely studied algorithms.

Data analytics on edge devices has gained rapid growth in research, industry, and different aspects of our daily life. This topic still faces many challenges such as limited computation resource on edge devices. In this paper, we further identify two main challenges: the composition and deployment of data analytics services on edge devices. We present the Zoo system to address these two challenge: on one hand, it provides simple and concise domain-specific language to enable easy and and type-safe composition of different data analytics services; on the other, it utilises multiple deployment backends, including Docker container, JavaScript, and MirageOS, to accommodate the heterogeneous edge deployment environment. We show the expressiveness of Zoo with a use case, and thoroughly compare the performance of different deployment backends in evaluation.

Internet-connected devices are increasingly present in our homes, and privacy breaches, data thefts, and security threats are becoming commonplace. In order to avoid these, we must first understand the behaviour of these devices. In this work, we analyse network traces from a testbed of common IoT devices, and describe general methods for fingerprinting their behavior. We then use the information and insights derived from this data to assess where privacy and security risks manifest themselves, as well as how device behavior affects bandwidth. We demonstrate simple measures that circumvent attempts at securing devices and protecting privacy., Comment: Submitted to TMA 2018

Machine Learning (ML) techniques, such as Neural Network, are widely used in today's applications. However, there is still a big gap between the current ML systems and users' requirements. ML systems focus on improving the performance of models in training, while individual users cares more about response time and expressiveness of the tool. Many existing research and product begin to move computation towards edge devices. Based on the numerical computing system Owl, we propose to build the Zoo system to support construction, compose, and deployment of ML models on edge and local devices.

Most machine learning and deep neural network algorithms rely on certain iterative algorithms to optimise their utility/cost functions, e.g. Stochastic Gradient Descent. In distributed learning, the networked nodes have to work collaboratively to update the model parameters, and the way how they proceed is referred to as synchronous parallel design (or barrier control). Synchronous parallel protocol is the building block of any distributed learning framework, and its design has direct impact on the performance and scalability of the system. In this paper, we propose a new barrier control technique - Probabilistic Synchronous Parallel (PSP). Com- paring to the previous Bulk Synchronous Parallel (BSP), Stale Synchronous Parallel (SSP), and (Asynchronous Parallel) ASP, the proposed solution e ectively improves both the convergence speed and the scalability of the SGD algorithm by introducing a sampling primitive into the system. Moreover, we also show that the sampling primitive can be applied atop of the existing barrier control mechanisms to derive fully distributed PSP-based synchronous parallel. We not only provide a thorough theoretical analysis1 on the convergence of PSP-based SGD algorithm, but also implement a full-featured distributed learning framework called Actor and perform intensive evaluation atop of it.

The capacity and programmability of reconfigurable hardware such as FPGAs has improved steadily over the years, but they do not readily provide any mechanisms for monitoring or debugging running programs. Such mechanisms need to be written into the program itself. This is done using ad hoc methods and primitive tools when compared to CPU programming. This complicates the programming and debugging of reconfigurable hardware. We introduce Program-hosted Directability (PhD), the extension of programs to interpret direction commands at runtime to enable debugging, monitoring and profiling. Normally in hardware development such features are fixed at compile time. We present a language of directing commands, specify its semantics in terms of a simple controller that is embedded with programs, and implement a prototype for directing network programs running in hardware. We show that this approach affords significant flexibility with low impact on hardware utilisation and performance.

Many current Internet services rely on inferences from models trained on user data. Commonly, both the training and inference tasks are carried out using cloud resources fed by personal data collected at scale from users. Holding and using such large collections of personal data in the cloud creates privacy risks to the data subjects, but is currently required for users to benefit from such services. We explore how to provide for model training and inference in a system where computation is pushed to the data in preference to moving data to the cloud, obviating many current privacy risks. Specifically, we take an initial model learnt from a small set of users and retrain it locally using data from a single user. We evaluate on two tasks: one supervised learning task, using a neural network to recognise users' current activity from accelerometer traces; and one unsupervised learning task, identifying topics in a large set of documents. In both cases the accuracy is improved. We also analyse the robustness of our approach against adversarial attacks, as well as its feasibility by presenting a performance evaluation on a representative resource-constrained device (a Raspberry Pi).