| Monday, April 29 | |
| 8:30 ‑ 9:00 | Opening Session: Keynote |
| 9:00 ‑ 10:00 | Session 1: Service Level Flexibility |
| 10:00 ‑ 10:30 | |
| 10:30 ‑ 12:30 | Session 2: Network Level Flexibility |
Room Studio A
Monday, April 29
Monday, April 29 8:30 – 9:00
Opening Session: Keynote
Room: Studio A
Chair: Ertan Onur (Middle East Technical University, Turkey)
Keynote Talk by Dr. Thrasyvoulos Spyropoulos, Associate Professor, EURECOM, France
Title: From Content-Centric to Entertainment-Centric Networks: Convergence of Communication, Caching, and Recommendation Systems at the Network Edge
Abstract: It is well known that there is a dramatic tussle between the users’ media consumption needs and what wireless operators are able to offer. Distributed storage of frequently requested content as close to the user as possible (e.g. at or near small cells or even in user devices) is considered as a “sine qua non” towards keeping up with user demand while controlling operator costs. While this has led to a spur of research around caching for wireless networks, a number of crucial shortcomings remain towards providing better user QoE for everyone, everywhere, and at an affordable cost. In this talk, we will describe how network flexibility and edge computing, will facilitate the convergence of the communication world (how content should be transmitted), the caching world (where content should be placed), and the application world (what content should be recommended), creating a software-based paradigm shift towards entertainment-centric networks. We will also discuss how modern optimization theory and machine learning could help tackle the hard problems this convergence gives rise to.
Short Bio: Thrasyvoulos Spyropoulos (male) is an Associate Professor at EURECOM, France since October 2010. He holds a Ph.D. degree from the University of Southern California (USC), Los Angeles, US, and before joining EURECOM he spent a year as a post-doctoral researcher at INRIA, Sophia-Antipolis, and 3 years as a Senior Researcher and Lecturer at ETH Zurich. His research interests include performance analysis, content-centric networks, cellular system modeling and optimization, mobile data offloading, mobility modeling, and social networks. He has co-authored more than 80 publications in international conferences and journals, receiving more than 10000 citations, has served in the TPC of numerous top-tier conferences, such as ACM Mobihoc, ACM Sigmetrics, and IEEE Infocom, and has co-chaired the ACM CHANTS 2013 and IEEE NetSciCom 2014 workshops. He is also the co-recipient of the best paper awards at IEEE Secon 2008, IEEE WoWMoM 2012, and the best paper award runner-up for ACM Mobihoc 2011.
Monday, April 29 9:00 – 10:00
Session 1: Service Level Flexibility
Room: Studio A
Chair: Mehmet Yunus Dönmez (Ericsson & Bogazici University, Turkey)
- A Market-based Modular Service Composition Approach for Flexibility and Adaptability in Future Wireless Networks
-
Online end-user service demand, supported by the convergence of next generation networks, has undergone a vast transformation with respect to heterogeneity and diversity. Networks and its elements are required to provide agile and flexible services in different user groups or market segments, towards realizing future network user-centric prospect. However, conventionally this happens only as a combination of existing and often competing service options. In this article, we discuss a novel paradigm based on the concept of service bundling, which structurally integrates various technical service modules based on customer preferences in order to optimize resource allocation and increase user overall experience in future wireless networks. Stemming from the commonalities and synergies among the bundled options in terms of characteristics and performance, a new and superior product is devised, enabling users to enjoy better performance under a common and standardized methodology. A modular approach is realized, which promotes adaptability and flexibility in the offering, while at the same time allows for standardization and scalability potentials.
- Enhancing QoE for Video Streaming Considering Congestion: A Fault Tolerance Approach
-
Since tremendous amount of traffic is generated in modern networks as a result of the mobility and video streaming, the congestion issue is faced more frequently in the networks. Accordingly, failures and performance losses in networks due to congestion result in deteriorated Quality of Experience (QoE) from the end user perspective that may cause financial and reputation loss for the service provider. Even though the new video streaming paradigm, Dynamic Adaptive Streaming over HTTP (DASH), is proposed as a solution for the changing condition of the networks, it is not sufficient considering the heavily loaded links that show the symptoms of link failures. Therefore, the flexible implementation of the data plane fault tolerance scheme that can be applied for other problems like congestion in networks is crucial. Thus, in this study, we apply the data plane fault tolerance approach in the Software-Defined Network to improve the QoE of DASH clients in the case of congestion rather than the failure. To detect the congestion in the network level, we use the Bidirectional Forwarding Protocol (BFD) that is originally implemented for link failures. In our experiments, we investigate the effect of the BFD interval, video segment size, and traffic load on QoE parameters. Our results show that if the fault tolerance approach is applied using a small BFD interval with a large segment size, QoE parameters are noticeably enhanced considering the non-applied case.
- Deep Reinforcement Learning for Dynamic Network Slicing in IEEE 802.11 Networks
-
Network slicing, a key enabler for future wireless networks, divides a physical network into multiple logical networks that can be dynamically created and configured. In current IEEE 802.11 (Wi-Fi) networks, the only form of network configuration is a rule-based optimization of few parameters. Future access points (APs) are expected to have self-organizational capabilities, able to deal with large configuration spaces in order to dynamically configure each slice. Deep Reinforcement Learning (DRL) can achieve promising results in highly dynamic and complex environments without the need for an operating model, by learning the optimal strategy after interacting with the environment. However, since the number of possible slice configurations is huge, achieving the optimal strategy requires an exhaustive learning period that might yield an outdated slice configuration. In this paper, we propose a fast-learning DRL model that can dynamically optimize the slice configuration of unplanned Wi-Fi networks without expert knowledge. Enhanced with an off-line learning step, the proposed approach is able to achieve the optimal slice configuration with a fast convergence, which is attractive for dynamic scenarios.
- Coffee Break 10:00-10:30
Monday, April 29 10:30 – 12:30
Session 2: Network Level Flexibility
Room: Studio A
Chair: Ilker Demirkol (Universitat Politecnica de Catalunya, Spain)
- Gossip-based monitoring of virtualized resources in 5G networks
-
The service function (SFs) area has gained increasing attention in the last years due its ability to combine the advantages of cloud computing and software defined networking (SDN) into the novel paradigm of network function virtualization (NFV). By decoupling SFs from the physical equipment where they are executed, NFV should allow increasing the efficiency of resource utilization, and make networks and services more scalable and flexible. However, in order to efficiently manage and chain SFs to build network service slices in 5G networks, it is necessary to localize (virtualized) SFs together with their current status, which includes their computing load, load of associated virtual links, configuration parameters, etc. To this aim, we propose a monitoring architecture able to track the network location and the current status of distributed and virtualized SFs, by using monitoring agents responsible to monitor the status of co-located SFs, both physical and virtual ones. The monitoring agents exchange their information by means of a gossip protocol, so as to increase the reliability of the process and to build a distributed service monitoring architecture. In this way, it is possible to keep service decisions as local as possible, limiting the interactions with a centralized orchestrator, and thus increasing network scalability. We show that the network overhead of the distributed monitoring process is definitely negligible.
- Towards Automatic Deployment of Virtual Firewalls to Support Secure mMTC in 5G Networks
-
Internet of Things (IoT) has emerged as the main enabler for dealing with challenging use cases that require massive Machine-Type Communications (mMTC), which has been recognized as one of three use case types for the Fifth Generation (5G) and beyond networks. In IoT networks, it is prohibitive to rely on just one firewall where hundreds of thousands of rules need to be installed in order to provide security counter-measures to each of the IoT devices. To fill this gap, this paper proposes an automatic deployment of virtual firewalls by leveraging Network Function Virtualisation (NFV) Management and Orchestration (MANO) to protect NB-IoT mMTC communications. The main idea underneath is to use NFV to deal with efficient rule distribution across VNFs-based firewalls to achieve scalability in the number of managed IoT devices. Empirical results have validated the design and implementation of the proposed scheme and demonstrating its advantageous performance and scalability. In particular, the deployment time for this VNF-based firewall service is highlighted to meet the requirement of a 5G Key Performance Indicator (KPI).
- Flexible and Efficient Deployment of NB-IoT and LTE-MTC in Coexistence with 5G New Radio
-
This paper addresses the coexistence of Narrowband IoT (NB-IoT) or Long Term Evolution for Machine-Type Communications (LTE-MTC, a.k.a. LTE-M) on one hand, and the 5th generation (5G) New Radio access (NR) on the other. We propose an effective scheme which maintains resource efficiency and prevents mutual interference, by optimally placing an NB-IoT or LTE-M carrier inside an NR carrier. Our scheme enables the deployment on the same spectrum of NR for mobile broadband and ultra reliable low-latency communications (URLLC) services and NB-IoT or LTE-M for IoT services. We analyze the problem of subcarrier grid and resource block alignments in the coexistence of NR with NB-IoT and LTE-M carriers. Finally, we provide comprehensive results on the optimal locations of NB- IoT and LTE-M which are deployed inside NR for various NR channel bandwidths.
- A Novel Behavioral Social-Aware D2D User Association Scheme based on Self-Propelled Voronoi
-
A Device-to-Device (D2D) wireless communication framework, in which devices are linked together directly rather than relying on fixed infrastructure to relay communication, can be envisioned as a social network. Similar to evolving society, a D2D network tends to be affected by selfishness among socially popular nodes resulting from the survivalist nature found in interaction games – in game or evolution theory. During evolution, participants accumulate obvious negative traits, i.e. selfishness or greediness. Addressing these negative traits among social nodes, which have a higher number of social connections, can be complex due to repeated central intervention. Central intervention here means central authoritative node or in this case central control station in macro cell network. To address this limitation, in this paper the authors have proposed a reactive p0 measure central node selection scheme to achieve low complexity device-centric D2D user association, to achieve results comparable to social-aware algorithms, without the associated complexity. The results show an overall improvement by a factor of 1.2 to 3.3, compared to conventional algorithms, by introducing the fφ measure as reactive procedure to work against selfish nodes compromising network performance. To define the traits of user nodes the paper brings together the mobility concept of Self-Propelled Voronoi (SPV) model and the selfish-altruistic nature of human evolution.
- Model-Based optimization for JT CoMP in C-RAN
-
In this paper, we develop a model-based optimization technique for JT CoMP (Joint transmission Coordinated Multi-Points) in the context of Cloud based Radio Access (C-RAN) and 5G networks. The objective is to ease decisions on the distribution of resources between CoMP and non-CoMP users according to the traffic load. We propose a mathematical queuing model which parameters are computed from real radio conditions that we reproduced in a cluster of cells. We use a closed form formula that fits the behavior of the considered scheme and allows the analysis of the cell throughput according to traffic load, radio conditions and the distribution of available resources among the CoMP and non-CoMP users within the cooperative macro cells. Moreover, the computation speed of the results allows us to develop a complex optimization model. The relevance of our work is that it can be integrated as a resource management tools to decide quickly on the distribution of the available resources among the CoMP and non-CoMP users of the cooperative macro cells to achieve the optimal capacity of the cells in terms of user throughput.
- An adaptive functional split in 5G networks
-
The 5G radio access network (RAN) features a partially centralized architecture, in which a subset of the network functions are deployed in a centralized unit. Centralizing these functions reduces operating costs and enables coordination techniques. However, the more functions are centralized, the more capacity is needed on the fronthaul network connecting centralized and distributed units. In addition, the required fronthaul capacity also depends on the instantaneous user traffic, which varies over time. Therefore, in order to optimize its performance, the 5G RAN should be able to dynamically adapt its centralization level to the user traffic. In this paper, we present the design of an adaptive RAN that can switch between two different centralization options at runtime. We provide design objectives and challenges, as well as measurement results from a working implementation.