Serverless computing is a popular cloud computing paradigm, which requires low response latency to handle on-demand user requests. There are two prominent techniques employed for reducing the response latency: keep fully initialized containers alive (Warm Container) or reduce the new container startup (cold start) latency. This paper presents the 3rd container startup mode: Hibernate Container, which starts faster than the cold start container mode and consumes less memory than the Warm Container mode. Hibernate Container is essentially a "deflated" Warm Container. Its application memory is swapped out to disk, the freed memory is reclaimed and file based mmap memory is cleaned-up. The Hibernate Container's deflated memory is inflated in response to user requests. As Hibernate Container's application is fully initialized, its response latency is less than the cold start mode; and as the application memory is deflated, its memory consumption is less than the Warm Container mode. Additionally, when a Hibernate Container is "woken up" to process a request, the Woken-up Container has similar response latency to Warm Container but less memory consumption because not all the deflated memory needs to be inflated. We implemented the Hibernate technique as part of the open source Quark secure container runtime project and our test demonstrated that Hibernate Container consumes about 7\% to 25\% of the Warm Container memory. All of this results in a higher deployment density, lower latency and appreciable improvements in the overall system performance.
相關內容
Tor is the most popular anonymous communication overlay network which hides clients' identities from servers by passing packets through multiple relays. To provide anonymity to both clients and servers, Tor onion services were introduced by increasing the number of relays between a client and a server. Because of the limited bandwidth of Tor relays, large numbers of users, and multiple layers of encryption at relays, onion services suffer from high end-to-end latency and low data transfer rates, which degrade user experiences, making onion services unsuitable for latency-sensitive applications. In this paper, we present a UDP-based framework, called DarkHorse, that improves the end-to-end latency and the data transfer overhead of Tor onion services by exploiting the connectionless nature of UDP. Our evaluation results demonstrate that DarkHorse is up to 3.62x faster than regular TCP-based Tor onion services and reduces the Tor network overhead by up to 47%.
Open and permissionless blockchains are distributed systems with thousands to tens of thousands of nodes, establishing novel platforms for decentralized applications. When realizing such an application, data might be stored and retrieved from one or more blockchains by distributed network nodes without relying on centralized coordination and trusted third parties. Data access could be provided through a query language such as SQL at the application level, establishing a unified view on application-level data that is verifiably stored. However, when accessing multiple blockchains through their node software and APIs, interoperability cannot be assumed today, resulting in challenges of inhomogeneous data access. In addition, different feature sets and trade-offs exist, e.g., regarding smart contract functionality, availability, distribution, scalability, and security. For increasing interoperability, the paper at hand suggests pursuing the development of a cross-chain query language at the application level. The language abstracts from implementation by providing a standardized syntax, an integrated data model, and a processing architecture for data queries. This research is an extended and updated paper demonstrating the language syntax, data model, and architecture with an evaluation of compatibility against the largest open and permissionless blockchains today.
In this paper, we present a robust deep incremental learning model for regression tasks on financial temporal tabular datasets. Using commonly available tabular and time-series prediction models as building blocks, a machine-learning model is built incrementally to adapt to distributional shifts in data. Using the concept of self-similarity, the model uses only a basic building block of machine learning methods, decision trees to build models of any required complexity. The model is demonstrated to have robust performances under adverse situations such as regime changes, fat-tailed distributions and low signal-to-noise ratios which is common in financial datasets. Model robustness are studied under different hyper-parameters such as model complexity and data sampling settings using XGBoost models trained on the Numerai dataset as a detailed case study. The two layer deep ensemble of XGBoost models over different model snapshots is demonstrated to deliver high quality predictions under different market regimes. Comparing the XGBoost models with different number of boosting rounds in three scenarios (small, standard and large), we demonstrated the model performances are monotonic increasing with respect to model sizes and converges towards the generalisation upper bound. Our model is efficient with much lower hardware requirement than other machine learning models as no specialised neural architectures are used and each base model can be independently trained in parallel.
Mobile Edge Computing (MEC) is expected to play a significant role in the development of 6G networks, as new applications such as cooperative driving and eXtended Reality (XR) require both communication and computational resources from the network edge. However, the limited capabilities of edge servers may be strained to perform complex computational tasks within strict latency bounds for multiple clients. In these contexts, both maintaining a low average Age of Information (AoI) and guaranteeing a low Peak AoI (PAoI) even in the worst case may have significant user experience and safety implications. In this work, we investigate a theoretical model of a MEC server, deriving the expected AoI and the PAoI and latency distributions under the First In First Out (FIFO) and Generalized Processor Sharing (GPS) resource allocation policies. We consider both synchronized and unsynchronized systems, and draw insights on the robust design of resource allocation policies from the analytical results.
Graph Neural Networks (GNNs) have gained growing interest in miscellaneous applications owing to their outstanding ability in extracting latent representation on graph structures. To render GNN-based service for IoT-driven smart applications, traditional model serving paradigms usually resort to the cloud by fully uploading geo-distributed input data to remote datacenters. However, our empirical measurements reveal the significant communication overhead of such cloud-based serving and highlight the profound potential in applying the emerging fog computing. To maximize the architectural benefits brought by fog computing, in this paper, we present Fograph, a novel distributed real-time GNN inference framework that leverages diverse and dynamic resources of multiple fog nodes in proximity to IoT data sources. By introducing heterogeneity-aware execution planning and GNN-specific compression techniques, Fograph tailors its design to well accommodate the unique characteristics of GNN serving in fog environments. Prototype-based evaluation and case study demonstrate that Fograph significantly outperforms the state-of-the-art cloud serving and fog deployment by up to 5.39x execution speedup and 6.84x throughput improvement.
Large Language Models work quite well with general-purpose data and many tasks in Natural Language Processing. However, they show several limitations when used for a task such as domain-specific abstractive text summarization. This paper identifies three of those limitations as research problems in the context of abstractive text summarization: 1) Quadratic complexity of transformer-based models with respect to the input text length; 2) Model Hallucination, which is a model's ability to generate factually incorrect text; and 3) Domain Shift, which happens when the distribution of the model's training and test corpus is not the same. Along with a discussion of the open research questions, this paper also provides an assessment of existing state-of-the-art techniques relevant to domain-specific text summarization to address the research gaps.
Bike sharing systems have been widely deployed around the world in recent years. A core problem in such systems is to reposition the bikes so that the distribution of bike supply is reshaped to better match the dynamic bike demand. When the bike-sharing company or platform is able to predict the revenue of each reposition task based on historic data, an additional constraint is to cap the payment for each task below its predicted revenue. In this paper, we propose an incentive mechanism called {\em TruPreTar} to incentivize users to park bicycles at locations desired by the platform toward rebalancing supply and demand. TruPreTar possesses four important economic and computational properties such as truthfulness and budget feasibility. Furthermore, we prove that even when the payment budget is tight, the total revenue still exceeds or equals the budget. Otherwise, TruPreTar achieves 2-approximation as compared to the optimal (revenue-maximizing) solution, which is close to the lower bound of at least $\sqrt{2}$ that we also prove. Using an industrial dataset obtained from a large bike-sharing company, our experiments show that TruPreTar is effective in rebalancing bike supply and demand and, as a result, generates high revenue that outperforms several benchmark mechanisms.
Motivated by distribution problems arising in the supply chain of Haleon, we investigate a discrete optimization problem that we call the "container delivery scheduling problem". The problem models a supplier dispatching ordered products with shipping containers from manufacturing sites to distribution centers, where orders are collected by the buyers at agreed due times. The supplier may expedite or delay item deliveries to reduce transshipment costs at the price of increasing inventory costs, as measured by the number of containers and distribution center storage/backlog costs, respectively. The goal is to compute a delivery schedule attaining good trade-offs between the two. This container delivery scheduling problem is a temporal variant of classic bin packing problems, where the item sizes are not fixed, but depend on the item due times and delivery times. An approach for solving the problem should specify a batching policy for container consolidation and a scheduling policy for deciding when each container should be delivered. Based on the available item due times, we develop algorithms with sequential and nested batching policies as well as on-time and delay-tolerant scheduling policies. We elaborate on the problem's hardness and substantiate the proposed algorithms with positive and negative approximation bounds, including the derivation of an algorithm achieving an asymptotically tight 2-approximation ratio.
The world population is anticipated to increase by close to 2 billion by 2050 causing a rapid escalation of food demand. A recent projection shows that the world is lagging behind accomplishing the "Zero Hunger" goal, in spite of some advancements. Socio-economic and well being fallout will affect the food security. Vulnerable groups of people will suffer malnutrition. To cater to the needs of the increasing population, the agricultural industry needs to be modernized, become smart, and automated. Traditional agriculture can be remade to efficient, sustainable, eco-friendly smart agriculture by adopting existing technologies. In this survey paper the authors present the applications, technological trends, available datasets, networking options, and challenges in smart agriculture. How Agro Cyber Physical Systems are built upon the Internet-of-Agro-Things is discussed through various application fields. Agriculture 4.0 is also discussed as a whole. We focus on the technologies, such as Artificial Intelligence (AI) and Machine Learning (ML) which support the automation, along with the Distributed Ledger Technology (DLT) which provides data integrity and security. After an in-depth study of different architectures, we also present a smart agriculture framework which relies on the location of data processing. We have divided open research problems of smart agriculture as future research work in two groups - from a technological perspective and from a networking perspective. AI, ML, the blockchain as a DLT, and Physical Unclonable Functions (PUF) based hardware security fall under the technology group, whereas any network related attacks, fake data injection and similar threats fall under the network research problem group.
In recent years, mobile devices have gained increasingly development with stronger computation capability and larger storage. Some of the computation-intensive machine learning and deep learning tasks can now be run on mobile devices. To take advantage of the resources available on mobile devices and preserve users' privacy, the idea of mobile distributed machine learning is proposed. It uses local hardware resources and local data to solve machine learning sub-problems on mobile devices, and only uploads computation results instead of original data to contribute to the optimization of the global model. This architecture can not only relieve computation and storage burden on servers, but also protect the users' sensitive information. Another benefit is the bandwidth reduction, as various kinds of local data can now participate in the training process without being uploaded to the server. In this paper, we provide a comprehensive survey on recent studies of mobile distributed machine learning. We survey a number of widely-used mobile distributed machine learning methods. We also present an in-depth discussion on the challenges and future directions in this area. We believe that this survey can demonstrate a clear overview of mobile distributed machine learning and provide guidelines on applying mobile distributed machine learning to real applications.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191