IMDEA Networks Institute Digital Repository
Not a member yet
1915 research outputs found
Sort by
Approximate Classifiers with Controlled Accuracy
No full textPerforming exact computations can require significant resources. Approximate computing allows to alleviate
resource constraints, sacrificing the accuracy of results. In this work, we consider a generalization of the classical packet classification problem. Our major contribution is to introduce various representations for approximate packet classifiers with controlled accuracy and optimization techniques to reduce classifier sizes exploiting this new level of flexibility. We validate our theoretical results with a comprehensive evaluation study.TRUEpu
Managing Hardware Impairments in Hybrid Millimeter Wave MIMO Systems: A Dictionary Learning-based Approach
Get PDFCompressed sensing-based strategies have been derived in prior work to reduce training overhead when estimating the high dimensional millimeter wave MIMO channel. These techniques rely on a channel model based on a sparsifying dictionary which does not account for hardware impairments such as calibration errors, mutual coupling effects, or manufacturing errors in the inter-spacing between the array elements. In this paper, we propose a learning strategy for the sparsifying dictionary that considers a channel model with hardware impairments, embedding these effects into the dictionary itself. This way, a sparser representation of the channel can be obtained even when considering realistic implementations of the antenna array and the radio frequency chains. Numerical simulations with different system configurations and parameters of the hardware impairments, show the effectiveness of the proposed dictionary learning algorithm for channel estimation at millimeter wave frequencies with hybrid MIMO architectures.TRUEpu
Miner Dynamics on the Ethereum Blockchain
Get PDFThe recent rise of Bitcoin, and blockchain–its underlying technology–has ushered in a wave of other cryptocurrencies and blockchain applications. One such application is the Ethereum, a decentralized software platform that rose to prominence for enabling smart contracts and distributed applications. The Ethereum blockchain is powered by Ether, a platform-specific token used for transactions and paying for services on the network. Like in Bitcoin, miners try to solve cryptographic puzzles in order to get rewards, in this case, Ethers. However, experience has shown that the returns on solo mining have a high variance i.e. returns are unpredictable. Therefore, miners come together to form mining pools with power equivalent to the sum of the power of each member miner. This guarantees more predictable earnings, albeit likely lower, since earnings are shared among miners in the pool according to an agreed scheme.
Miners are allowed to migrate between mining pools as they deem fit which has led to a lot of dynamics between pools. This raises the question about what factors influence miner dynamics, like migrations. Can we understand these factors and their effect on the decentralization of the system? Are miners leaving, joining and changing pools mainly to maximize earnings and returns on their investments (on computing power)? Are there colluding miners or pools? As part of our efforts to answer some of these questions, we obtained data from the Ethereum blockchain system
with records of about the 60 most prominent pools in the system (which account for more than 90% of the power), their miners, reward payouts, and miner movements including those that join, leave, or switch pools. These records are broken down into time windows each of four weeks.
Initial analysis seems to suggest a preferential attachment and detachment (when miners decide
to join or leave a pool) with fairly strong correlations between pool power and the percentage of
miners migrating from and to other pools across all time windows. Furthermore, distributions of
relative miner power (inferred from the percentage of payouts they get) in some time windows seem
to follow power-law distribution while over other time windows follow a lognormal distribution.
To understand miner migrations between pools, we build a network from the payouts records with
pools as nodes and miners as edges. Hence, two pools i and j are connected with an edge of weight
n if n miners have received a payout from both pools i and j at any point in time. This network
revealed a few number of well-connected pools with lots of common miners between them. This
may indicate that these pools are the major players in the system, but a lot of common miners may
also indicate colluding or related pools, i.e. pools that are seemingly different but are operated by the same entity. Using four different network clustering/community detection methods, namely fast-greedy, walktrap, edge betweenness, and leading eigenvalues, reveal a community of about 30 pools consistently identified across 3 of the 4 community detection methods. Furthermore, pools that are known to be operated by the same entity are usually clustered in the same community.
As a continuation of this work, we expect to come up with a model that adequately explains the miner dynamics, miner attachments and detachments i.e. how they choose pools to join or leave. Furthermore, we hope to continue to explore the network aspects of this work to gain more
insights into the Ethereum ecosystem.FALSEpu
On the Feasibility of Using IEEE 802.11ad mmWave for Accurate Object Detection
Get PDFMillimeter wave (mmWave) bands are considered highly for localization and object detection. In this paper we assess the potential of commercial IEEE 802.11ad mmWave equipment to offer accurate object detection, ultimately providing models of the physical environment. Unlike solutions using bespoke mmWave equipment for detection, the use of IEEE 802.11ad ensures a low-cost system, and one in which detection can be integrated with communication, creating potential for innovative applications. Our approach is to build a laboratory testbed in which we capture reflected mmWave signals that are generated and transmitted by a commercial offthe-shelf (COTS) IEEE 802.11ad mmWave device. From the measured channel impulse response, we measured the distance from the mmWave transceiver to the objects in the environment, by some simple signal processing techniques. By knowing the angle of mmWave departure/arrival and this measured distance, we can develop a 2D model of the physical environment. We report on the achieved accuracy, which is 2cm in most experiments, and discuss technology limitations and research opportunities.TRUEpu
Machine Learning Methods for Reliable Resource Provisioning in Edge-Cloud Computing: A Survey
Get PDFLarge-scale software systems are currently designed as distributed entities and deployed in cloud data centers. To overcome the limitations inherent to this type of deployment, applications are increasingly being supplemented with components instantiated closer to the edges of networks—a paradigm known as edge computing. The problem of how to efficiently orchestrate combined edge-cloud applications is, however, incompletely understood, and a wide range of techniques for resource and application management are currently in use.
This article investigates the problem of reliable resource provisioning in joint edge-cloud environments, and surveys technologies, mechanisms, and methods that can be used to improve the reliability of distributed applications in diverse and heterogeneous network environments. Due to the complexity of the problem, special emphasis is placed on solutions to the characterization, management, and control of complex distributed applications using machine learning approaches. The survey is structured around a decomposition of the reliable resource provisioning problem into three categories of techniques: workload characterization and prediction, component placement and system consolidation, and application elasticity and remediation. Survey results are presented along with a problem-oriented discussion of the state-of-the-art. A summary of identified challenges and an outline of future research directions are presented to conclude the article.pu
Performance Assessment of Off-the-Shelf mmWave Radios for Drone Communications
Get PDFThis paper presents experiments to assess and understand the feasibility of millimeter-wave (mmWave) radios for aerial links at low altitude, namely drone communications. In this preliminary study, a Commercial Off-The-Shelf (COTS) mmWave radio is attached to a DJI Matrice 600 Pro drone. The measurement campaign reveals that while the technology is promising, the range is extremely limited, i.e., approximately 30m. Our analysis shows that the poor range is likely due to low-directionality of today's COTS devices as well as poor rate adaptation in mobile environments. More antennas at both the transmitter and the receiver, along with better 3D beamforming, will be essential to boosting the communication range and thus making this a viable technology for use cases such as high definition real-time monitoring in disaster response scenarios or rapid deployment of multi-Gbps aerial links to expand connectivity in underserved areas.TRUEpu
Fair and Throughput-Optimal Routing in Multimodal Underwater Networks
Get PDFWhile acoustic communications are still considered the most prominent technology to communicate under water, other technologies are being developed based, e.g., on optical and radio-frequency electromagnetic waves. Each technology has its own advantages and drawbacks: for example, acoustic signals achieve long communication ranges at order-of-kbit/s bit rate, whereas optical signals offer order-of-Mbit/s transmission rates, but only over short ranges. Such a diversity can be leveraged by multimodal systems, which integrate different technologies and provide the intelligence required to decide which one should be used at any given time.
In this paper, we address a fundamental part of this intelligence by proposing Optimal Multimodal Routing (OMR), a novel routing protocol for underwater networks of multimodal nodes. OMR makes distributed decisions about the flow in each link and over each technology at any given time, in order to advance a packet towards its destination; in doing so, it prevents bottlenecks and allocates resources fairly to different nodes. We analyze the performance of OMR via simulations and in a field experiment. The results show that OMR successfully leverages all technologies to deliver data, even in the presence of imperfect topology information. To permit the reproduction of our results, we share our simulation code.pu
Why energy matters? Profiling energy consumption of mobile crowdsensing data collection frameworks
Get PDFMobile Crowdsensing (MCS) has emerged in the last years and has become one of the most prominent paradigms for urban sensing. The citizens actively participate in the sensing process by contributing data with their mobile devices. To produce data, citizens sustain costs, i.e., the energy consumed for sensing and reporting operations. Hence, devising energy efficient data collection frameworks (DCF) is essential to foster participation. In this work, we investigate from an energy-perspective the performance of different DCFs. Our methodology is as follows: (i) we developed an Android application that implements the DCFs, (ii) we profiled the energy and network performance with a power monitor and Wireshark, (iii) we included the obtained traces into CrowdSenSim simulator for large-scale evaluations in city-wide scenarios such as Luxembourg City, Turin and Washington DC. The amount of collected data, energy consumption and fairness are the performance indexes evaluated. The results unveil that DCFs with continuous data reporting are more energy-efficient and fair than DCFs with probabilistic reporting. The latter exhibit high variability of energy consumption, i.e., to produce the same amount of data, the associated energy cost of different users can vary significantly.pu
On the Scalability of Connectivity Services in a Multi-Operator Orchestrator Sandbox
Get PDFThe paper investigates the performance of a multi-domain orchestrator (MdO) deployed in a real multi-domain European testbed. Results show how the MdO prototype scales well with the number of domains advertised and connectivity services provisioned.TRUEpu
Panel: Smart connectivity in the context of Next Generation Internet
No full textPanel Organizer: Arturo Azcorra
Panel Members: Arturo Azcorra (5TONIC, Spain), Serge Fdida (Sorbonne University, France), Diego Lopez (Telefonica I+D, Spain), Werner Mohr (Nokia, Germany), Dimitra Simeonidou (University of Bristol, United Kingdom)FALSEpu