PHY-Aware Edge Caching in 5G: Convergence of Network and Information Theory in Cache-aided Wireless Networks
Nous vous rappelons que, afin de garantir l'accès de tous
les inscrits aux salles de réunion, l'inscription aux réunions est
gratuite mais obligatoire.
Inscriptions closes à cette réunion.
Inscriptions
4 personnes membres du GdR ISIS, et 16 personnes non membres du GdR, sont inscrits à cette réunion.
Capacité de la salle : 130 personnes.
Annonce
There is currently an urgent need for novel technologies that can partially mitigate the current explosion of wireless traffic volumes. Caching at the edge of the wireless network (e.g. on edge clouds, small cells, or even user devices) is considered as one of the major enabling technologies for 5G networks. This has revived the field of caching, and very promising initial results have been reported from both technical breakthroughs, such as coded caching and joint optimization of caching and transmission in network MIMO setups, as well as architectural breakthroughs such as the coordinated caching in many nearby small caches (femto-caching) or the convergence of content providers (e.g. Google, Netflix), CDN providers (e.g. AKAMAI) and ISPs.
The goal of this GdR ISIS meeting will be twofold. First, to present some of the state-of-the-art ideas considering caching "on top" of physical layer communications, offering an educational overview of the latest progress in designing algorithms that manage to handle multiuser interference by using cache-aided and feedback-aided MIMO techniques, to (often jointly) elevate performance, from that achieved by serving one user at a time (TD), to the much higher performance where many users are served simultaneously and seamlessly without interference. As a second goal, the GdR ISIS meeting will discuss some recent architectural changes like Mobile Edge Computing, RAN Sharing, and distributed nano-datacenters, facilitate this convergence between coordinated communication and cooperative caching, and allows the operator or the content provider (acting as a virtual operator) to directly and jointly optimize both, near the edge of the network. Finally, the speakers will be asked to share their vision on how the challenges and opportunities in each of these network-, communication-, and system-centric views of wireless caching might be in conflict, and how they could potentially be aligned.
Organisateurs
- Georgios Paschos (Huawei, Paris, georgios.paschos@huawei.com)
- Thrasyvoulos Spyropoulos (Eurecom, thrasyvoulos.spyropoulos@eurecom.fr)
- Petros Elia (Eurecom, elia@eurecom.fr)
Programme
Speakers:
-
- Osvaldo Simeone -- Plenary (Kings College London)
- James Roberts (Telecom ParisTech)
- Eleftherios Lampiris (EURECOM)
- Anastasios Giovannidis (UPMC - LIP6)
- Qifa Yan (Telecom ParisTech)
- Philippe Ciblat (Telecom ParisTech)
- Sheng Yang, (CentraleSupelec)
- Giovanni Neglia (INRIA, Sophia Antipolis)
- Mireille Sarkiss (CET)
- Luigi Vigneri (Huawei Research Labs, Paris)
Programme
===========================
8h30-9h00: Reception with Coffee
===========================
9h00-9h05: Introduction
9h05-9h50: Osvaldo Simeone - Plenary
Title: Fog-Aided Wireless Networks: An Information-Theoretic View
9h50-10h15: James Roberts
Title: Realizing an optimal memory-bandwidth tradeoff for content delivery in the access network
10h15-10h40: Eleftherios Lampiris
Title: Coded Caching: Discussing Promises and Challenges
===========================
10h40-11h10: Break
===========================
11h10-11h35: Anastasios Giovannidis
Title: Edge-caches for Rent
11h35-12h00: Qifa Yan
Title: Placement delivery array design----A new approach for coded caching schemes
12h00-12h25: Philippe Ciblat
Title: Age-Optimal Constrained Cache Updating
===========================
12h25 - 14h10: Lunch Break
===========================
14h10-14h35: Sheng Yang
Title: Scalable Wireless Content Delivery with Coded Caching
14h35-15h00: Giovanni Neglia
Title: Implicit Coordination of Caches in Small Cell Networks under Unknown Popularity Profiles
15h00-15h25: Mireille Sarkiss
Title: Coded Caching for Wiretap Broadcast Channels
15h25-15h50: Luigi Vigneri
Title: Vehicles as a Mobile Cloud
===========================
15h50-16h15: Coffee Break
===========================
16h15 - 17:30 Panel discussion
Résumés des contributions
Plenary speaker: Osvaldo Simeone
Title: Fog-Aided Wireless Networks: An Information-Theoretic View
Abstract: Fog-aided wireless networks are an emerging class of wireless systems that leverage the synergy and complementarity of cloudification and edge processing. The operation of fog-aided wireless networks requires the management of the communication, caching and computing resources at the cloud and at the edge, as well as to the transmission on the fronthaul network connecting cloud and edge. In this talk, this problem will be discussed in the context of content delivery from the point of view of information theory. Considering both offline and online caching, fundamental theoretical insights and algorithmic guidelines will be discussed. This work has received funding from the European Research Council (ERC) under the European Union?s Horizon 2020 research and innovation programme (grant agreement No. 725731).
==============================
Speaker: Jim Roberts
Title : Realizing an optimal memory-bandwidth tradeoff for content delivery in the access network
Abstract: We postulate an ICN architecture where content is mainly cached in central offices that take the form of a data center. Caches may also be placed at access nodes closer to users but not in core routers. We evaluate the effectiveness of caching under popularity churn and deduce the need for proactive caching particularly in access nodes and points closer to the end users (like wireless base stations). The optimal memory for bandwidth tradeoff is determined as a function of cost, demand and content popularity parameters. It appears difficult to realize this optimum in practice, however. While ISPs clearly seek to minimize their infrastructure investments, required content popularity data and control over proactive placement are jealously guarded by content providers. As the "subsidy side" of the content delivery 2-sided market, there is little economic incentive for the content providers to satisfy the ISPs' requirements.
==============================
Speaker: Eleftherios Lampiris
Title: Coded Caching: Discussing Promises and Challenges
Abstract: Coded Caching is a promising solution to significantly reduce network traffic related to delivery of pre-cached content. By intelligently placing part of the content at the users? caches in a pre-fetching phase, it allows, when the users? requests have been made, to serve many users at the same time by using unwanted cached content to remove interference.
In this talk, we will focus on discussing the mechanics of Coded Caching and some of the most promising extensions of it. Moreover, we will, also, look at some of the challenges that currently constrain Coded Caching results. Joint work with Petros Elia
==============================
Speaker: Anastasios Giovannidis
Title: Edge-caches for Rent
Abstract: Caching popular content at the wireless edge is recently proposed as a means to reduce congestion at the backbone of cellular networks. The two main actors involved are Mobile Network Operators (MNOs) and Content Providers (CPs). In this talk we will assume the following arrangement: an MNO pre-installs memory on its wireless equipment (e.g. Base Stations) and invites a unique CP to make use them, by renting them with some monetary cost. The CP will lease memory space and place its content; the MNO is responsible to associate network users to stations. The MNO could help (or not) the CP offload traffic from its data-center to the rented caches, depending on whether the association policy takes into account content placement.
A mixed-integer non-linear optimization problem from the CP perspective will be formulated, to maximize traffic offloading with minimum leasing costs. This is a joint optimization problem that can include any association policy, and can also derive the optimal one. A general exact solution using Benders? decomposition will be presented, which iteratively updates decisions of the two actors separately and converges to the global optimum. The optimal CP leasing/placement strategies and hit probability gains will be presented for various objectives and under different association policies. A main conclusion is that performance is maximised when the MNO association supports CP actions.
==============================
Speaker: Qifa Yan
Title: Placement delivery array design----A new approach for coded caching schemes
Abstract: Coded caching is a new technique to reduce the load of the backhaul link proposed by Maddah-Ali and Niesen, which receives significant interest in the literature. Recently, placement delivery array (PDA) is proposed to characterize both the placement and delivery phases of coded caching schemes with a single array. Moreover, given a PDA, a corresponding coded caching scheme can be implemented. Thus, through investigating the new constructions of PDA, it is possible to recover much more new coded caching schemes, some of which has some specific advantages in certain aspects. This talk will introduce the PDA framework and its connection to caching systems. Some relevant results and open problems will be briefly presented. (work with Xiaohu Tang, Minquan Cheng, and Qingchun Chen)
==============================
Speaker: Philippe Ciblat
Title : Age-Optimal Constrained Cache Updating
Abstract: We consider a system where a local cache maintains a collection of N dynamic content items that are randomly requested by local users. A capacity-constrained link to a remote network server limits the ability of the cache to hold the latest version of each item at all times, making it necessary to design an update policy. Using an age of information metric, we show under a relaxed problem formulation that an asymptotically optimal policy updates a cached item in proportion to the square root of the item's popularity. We then show experimentally that a physically realizable policy closely approximates the asymptotic optimal policy. Extension to time-varying popularities case will be evoked.
==============================
Speaker: Sheng Yang
Title: Scalable Wireless Content Delivery with Coded Caching
Abstract: In this talk, we consider a wireless network with a content server and K users each equipped with a cache. We show different scalable communication schemes with respect to the number of users, i.e., schemes that can allow the content delivery sum rate to scale linearly with K. Such schemes can be categorized into two classes. The first one harnesses the available diversity (space/time/frequency) in the system, where as the second one exploits limited feedback with simple scheduling.
==============================
Speaker: Giovanni Neglia
Title: Implicit Coordination of Caches in Small Cell Networks under Unknown Popularity Profiles
Abstract. In this talk we consider a dense cellular network, in which a limited-size cache is available at every Base Stations (BSs). In order to optimize the overall performance of the system in such scenario, where a significant fraction of the users is covered by several BSs, a tight coordination among nearby caches is needed.
To this end, we introduce a class of simple and fully distributed caching policies, which require neither direct communication among BSs, nor a priori knowledge of content popularity. Furthermore we propose a novel approximate analytical methodology to assess the performance of interacting caches under such policies. Our approach builds upon the well known characteristic time approximation and provides predictions that are surprisingly accurate (hardly distinguishable from the simulations) in most of the scenarios. Both synthetic and trace-driven results show that the our caching policies achieve excellent performance (in some cases provably optimal). They outperform state-of-the-art dynamic policies for interacting caches, and, in some cases, also the greedy content placement, which is known to be the best performing polynomial algorithm under static and perfectly-known content popularity profiles. (joint work with Emilio Leonardi, Politecnico di Torino, Italy)
==============================
Speaker: Mireille Sarkiss
Title: Coded Caching for Wiretap Broadcast Channels
Abstract: Caching has emerged as an important technique to increase rates of communication in networks that are highly loaded during certain periods. In this presentation, we show that caching naturally also serves to improve secrecy of communication. In fact, as we show, the gains of caching are much more pronounced when the communication xneeds to be kept secure from an external eavesdropper, than in the standard non-secure setup. In this presentation, we will describe various securing coding schemes that can be put in place thanks to the cache memories. We will compare the rates achieved by these schemes to our new information-theoretic upper bounds on the cache-aided secrecy rate. For simplicity of exposition, in the presentation we restrict to erasure channels, the results extend however to more general channels.
The presentation is based on joint work with S. Kamel, M. Wigger, G. Rekaya-Ben Othman.
==============================
Speaker: Luigi Vigneri
Title: Vehicles as a Mobile Cloud
Abstract: The large diffusion of handheld devices is leading to an exponential growth of the mobile traffic demand which is already overloading the core network. To deal with such a problem, several works suggest to store content in small cells or user equipments. In our work, we push the idea of caching at the edge a step further, and we propose to use public or private transportation as mobile small cells and caches. The adoption of such a mobile cloud, which does not suffer from energy constraints (compared to user equipments), reduces installation and maintenance costs (compared to small cells). In our framework, a user can opportunistically download chunks of a requested content from nearby vehicles, and be redirected to the cellular network after a deadline (imposed by the operator) or when her playout buffer empties. In this talk, we will suggest to an operator how to optimally replicate content to minimize the load on the core network, and we will show that, even with low technology penetration, the proposed caching policies are able to offload a large percentage of the mobile traffic demand. (Joint work with Thrasyvoulos Spyropoulos and Chadi Barakat)