IMDEA Networks Institute Digital Repository
Not a member yet
1915 research outputs found
Sort by
Paths and Interconnectivity: An Internet Trend and an Architectural Proposal
No full textThe Internet is a diverse ecosystem where a multitude of networks interconnect to provide end users with a global reach. Interconnection agreements supply economic incentives for networks to deliver traffic along end-to-end paths. Driven by the rapid Internet adoption, unrelenting traffic growth, and increasing demands for performance and quality, Internet interconnections evolve and affect traffic routing.
In this seminar, we first explore a phenomenon of remote peering, an interconnection where remote networks peer via a layer-2 provider. While our measurements reveal significant presence of remote peering at IXPs (Internet eXchange Points) worldwide and a substantial potential to offload transit traffic, we also discuss implications of remote peering for Internet topology modeling. Then, we present Route Bazaar, a new architecture for flexible Internet connectivity. Inspired by cryptocurrencies, the use of a blockchain in Route Bazaar enables multilateral contracts for end-to-end routing with QoS (Quality of Service), rich private policies, and public accountability.
The seminar is based on collaborations with Ignacio Castro, Camilo Cardona, Pierre Francois, Aurojit Panda, Barath Raghavan, and Scott Shenker.FALSEpu
VENUE: Virtualized Environment for multi-UAV network emulation
Get PDFUnmanned Aerial Vehicles (UAVs) have progressively been integrated into people lives during the last years. It is quite common now to see UAVs flying in the countryside doing field inspection, in highways for traffic control operations, or above stadiums in sport and music events. It is also common to see spectacular UAV swarm showcases (in most cases they are just performing a choreography) showing the potential of upcoming technologies. This article is focused on multi-UAV scenarios, on the establishment of Flying Ad hoc Networks (FANETs), and on the integration of 5G technologies like Network Function Virtualization (NFV) or Software Defined Networking (SDN). In particular, this article presents a proposal for one of the most common problems that the research and development community has to face at some stage: the validation of the different solutions and deployments. In this area, there is currently a notorious gap between the design phase and the deployment phase, since traditional network simulators are not designed with the constraints imposed by UAVs in mind. Besides, services implementations (that are usually distributed into single-board computers carried as payloads by UAVs) cannot be easily combined with the simulators. VENUE (Virtualized Environment for multi-UAV network emulation) is presented as an experimentation platform that allows testing the integration of multi-UAV FANETs together with network services deployments. VENUE covers from the simulation/emulation phase up to the real equipment integration phase. The validation of the platform is also presented in this article through several UAV use cases that make use of NFV technologies.pu
RL-Cache: Learning-Based Cache Admission for Content Delivery
No full textContent delivery networks (CDNs) distribute much of the Internet content by caching and serving the objects requested by users. A major goal of a CDN is to maximize the hit rates of its caches, thereby enabling faster content downloads to the users. Content caching involves two components: an admission algorithm to decide whether to cache an object and an eviction algorithm to decide which object to evict from the cache when it is full. In this paper, we focus on cache admission and propose a novel algorithm called RL-Cache that uses model-free reinforcement learning (RL) to decide whether or not to admit a requested object into the CDN's cache. Unlike prior approaches that use a small set of criteria for decision making, RL-Cache weights a large set of features that include the object size, recency, and frequency of access. We develop a publicly available implementation of RL-Cache and perform an evaluation using production traces for the image, video, and web traffic classes from Akamai's CDN. The evaluation shows that RL-Cache improves the hit rate in comparison with the state of the art and imposes only a modest resource overhead on the CDN servers. Further, RL-Cache is robust enough that it can be trained in one location and executed on request traces of the same or different traffic classes in other locations of the same geographic region.FALSEpu
Time-of-flight Wireless Indoor Navigation System for Industrial Environment
Get PDFWe introduce TWINS, Time-of-flight based Wireless Indoor Navigation
System, that estimates in near
real-time the position of commercial off-the-shelf (COTS)
devices equipped with WiFi.
TWINS uses COTS WiFi Access Points (APs)
and it does not require access to any inertial
sensors on the mobile devices.
We study the performance of TWINS in a harsh environment, strongly affected by the presence of metallic objects.
The scenario can be considered representative of an industrial indoor
environment with open spaces surrounded by metal obstacles.
We experimentally show that our system can localize four mobile robots tracked together,
with median error between 1.8 and 3.8 m.TRUEpu
Profiling Performance of Application Partitioning for Wearable Devices in Mobile Cloud and Fog Computing
Get PDFWearable devices have become essential in our daily activities. Due to battery constrains the use of computing, communication, and storage resources is limited. Mobile Cloud Computing (MCC) and the recently emerged Fog Computing (FC) paradigms unleash unprecedented opportunities to augment capabilities of wearables devices. Partitioning mobile applications and offloading computationally heavy tasks for execution to the cloud or edge of the network is the key. Offloading prolongs lifetime of the batteries and allows wearable devices to gain access to the rich and powerful set of computing and storage resources of the cloud/edge. In this paper, we experimentally evaluate and discuss rationale of application partitioning for MCC and FC. To experiment, we develop an Android-based application and benchmark energy and execution time performance of multiple partitioning scenarios. The results unveil architectural trade-offs that exist between the paradigms and devise guidelines for proper power management of service-centric Internet of Things (IoT) applications.pu
RL-NSB: Reinforcement Learning-Based5G Network Slice Broker
Get PDFNetwork slicing is considered one of the mainpillars of the upcoming 5G networks. Indeed, the ability toslice a mobile network and tailor each slice to the needs ofthe corresponding tenant is envisioned as a key enabler forthe design of future networks. However, this novel paradigmopens up to new challenges, such as isolation between networkslices, the allocation of resources across them, and the admissionof resource requests by network slice tenants. In this paper,we address this problem by designing the following buildingblocks for supporting network slicing: i) traffic and user mobil-ity analysis, ii) a learning and forecasting scheme per slice,iii) optimal admission control decisions based on spatial andtraffic information, and iv) a reinforcement process to drivethe system towards optimal states. In our framework, namelyRL-NSB, infrastructure providers perform admission controlconsidering the service level agreements (SLA) of the differenttenants as well as their traffic usage and user distribution, andenhance the overall process by the means of learning and thereinforcement techniques that consider heterogeneous mobilityand traffic models among diverse slices. Our results show that byrelying on appropriately tuned forecasting schemes, our approachprovides very substantial potential gains in terms of systemutilization while meeting the tenants’ SLAs.pu
Small Solar Panels Can Drastically Reduce the Carbon Footprint of Radio Access Networks
Get PDFThe limited power requirements of new generations of base stations make the use of renewable energy sources, solar in particular, extremely attractive for mobile network operators. Exploiting solar energy implies a reduction of the network operation cost as well as of the carbon footprint of radio access networks, but previous research works indicate that the area of the solar panels that are necessary to power a standard macro BS is large, so large to make the solar panel deployment problematic, especially within urban areas.
In this paper we use a modeling approach based on Markov reward processes to investigate the possibility of combining small area solar panels with a connection to the power grid to run a macro base station. By so doing, it is possible to increase the amount of renewable energy used to run a radio access network, while also reducing the cost incurred by the network operator to power its base stations. We assume that energy is drawn from the power grid only when needed to keep the BS operational, or during the night, that corresponds to the period with lowest electricity price. This has advantages in terms of both cost and carbon footprint.
We show that solar panels of the order of 1-2 kW peak, i.e., with a surface of about 5-10 square meters, combined with limited capacity energy storage (of the order of 10-15 kWh, corresponding to about 3-5 car batteries), and a smart energy management policy, can lead to an effective exploitation of renewable energy.TRUEpu
Unsupervised Scalable Statistical Method for Identifying Influential Users in Online Social Networks
Get PDFTRUEpu
Optimization of an integrated fronthaul/backhaul network under path and delay constraints
Get PDFCloud or Centralized Radio Access Networks (C-RANs) are
expected to be widely deployed under 5G in order to support the anticipated increased traffic demands and reduce costs. Under C-RAN, the radio elements (e.g., eNB or gNB in 5G) are split into a basic radio part (Distributed Unit, DU), and a pool-able base band processing part (Central Unit, CU). This func- tional split results in high bandwidth and delay constrained traffic flows between DUs and CUs (referred to as fronthaul), calling for the deployment of a specialized network to accommodate them or for integrating them with the rest of the flows (referred to as backhaul) over the existing infrastructure. This work studies the next generation of transport networks, which aims at integrating fronthaul and backhaul traffic over the same transport stratum. An optimization framework for routing and resource placement is developed, taking into account delay, capacity and path constraints, maximizing the degree of DU deployment while minimizing the supporting CUs. The frame- work and the developed heuristics (to reduce the computational complexity) are validated and applied to both small and large- scale (production-level) networks. They can be useful to network operators for both network plan- ning as well as network operation adjusting their (virtualized) infrastructure dynamically.pu