Elsevier

Computer Networks

Volume 93, Part 2, 24 December 2015, Pages 389-403
Computer Networks

Improving P2P streaming in Wireless Community Networks

https://doi.org/10.1016/j.comnet.2015.09.024Get rights and content

Abstract

Wireless Community Networks (WCNs) are bottom-up broadband networks empowering people with their on-line communication means. Too often, however, services tailored for their characteristics are missing, with the consequence that they have worse performance than what they could. We present here an adaptation of an Open Source P2P live streaming platform that works efficiently, and with good application-level quality, over WCNs. WCNs links are normally symmetric (unlike standard ADSL access), and a WCN topology is local and normally flat (contrary to the global Internet), so that the P2P overlay used for video distribution can be adapted to the underlaying network characteristics. We exploit this observation to derive overlay building strategies that make use of cross-layer information to reduce the impact of the P2P streaming on the WCN while maintaining good application performance. We experiment with a real application in real WCN nodes, both in the Community-Lab provided by the CONFINE EU Project and within an emulation framework based on Mininet, where we can build larger topologies and interact more efficiently with the mesh underlay, which is unfortunately not accessible in Community-Lab. The results show that, with the overlay building strategies proposed, the P2P streaming applications can reduce the load on the WCN to about one half, also equalizing the load on links. At the same time the delivery rate and delay of video chunks are practically unaffected.

Introduction

Wireless Community Networks (WCNs) are infrastructures built by the people for the people [1], with a bottom-up approach bringing broadband services to communities, rural and urban as well. They have become an important reality in the landscape of telecommunication and ICT services, and in some cases involve hundreds or thousands of nodes and people.1 They spread over large regions and some effort is currently spent in planning a sort of interconnecting federation [2]. The continuous growth of these networks and the services experimented on them represent a major swerve from the dominating paradigm of consolidation and centralization of the Internet. The increasing traffic generated by their users requires careful studies on the resources involved and the performance achievable by applications, with a special focus on services that have been ostracized in the commercial Internet. Among these, P2P applications, divested from their illegal aura, suite very well the WCN philosophy (delivering user-generated contents to other users) and also its internal, distributed structure, and can constitute killer-applications. Indeed, a P2P distributed architecture realized through a mesh overlay can exploit the WCN link characteristics like symmetric bandwidth and high throughput. On the contrary, applications based on the client-server approach cannot scale without a large amount of centralized resources because they do not take advantage from the underlaying network structure.

One of the P2P applications that raises most interest in a WCN is live streaming, which can have a relevant impact on the community. Live broadcasting of community user ideas and events fosters the opinion exchange within the community and can be a means for information propagation in a way much more tied to the citizens with respect to what traditional media do.

Even if the P2P architecture is distributed, the application must manage network resources accurately to avoid starvation and to properly balance the load in the underlying physical network. Our early works [3] and [4] report the feasibility of P2P live streaming in WCNs, but also highlight possible limitations affecting these kind of networks, and describe possible tuning needed to maximize the distributed content quality while limiting the network load.

This paper extends the work in [4], proposing and evaluating strategies for the management of the overlay topology and the content distribution in P2P live streaming. The key contributions of this paper can be summarized in the following bullets.

  • The proposed strategies are aware and respectful of the network resources in a WCN, and autonomously adapt to them to achieve optimal performance with minimal resource use and maximal fairness.

  • Several tests are performed on Community-Lab [2]. Community-Lab is a WCN testbed made up of real community network nodes, it allows the execution of experiments on different European WCNs testing the behavior of real-world application. Measures and consequent conclusions taken on Community-Lab are particularly relevant as proof of concept of the feasibility in real WCNs, and give high confidence on the final reception quality of the content distribution.

  • Additional tests have been run on Mininet [5]. Mininet is a network emulator we adapted to emulate actual WCNs. Our adaptation relays on the statistical data we collected from the Ninux WCN. This emulator allows the analysis of complex techniques like cross-layer optimization that are not feasible in Community-Lab.

The rest of the paper is structured as follows: Section 2 reviews the relevant literature and explains the goal of this paper; Section 3 describes P2P architecture and strategies and PeerStreamer, the platform we use for our tests; Section 4 is devoted to describe the strategies we propose to create an optimal overlay; Section 5 summarizes what are, at the application level, the strategies adopted to schedule video chunks transmissions; Section 6 presents the tools and infrastructures we use for testing our strategies; Section 7 reports the results we obtained from the tests and, finally, conclusions are given in Section 8.

Section snippets

State of the art and contribution of the paper

Cross-layer optimization has been shown to be a key factor for real-time content distribution in wireless mesh networks [6], [7]. However, the approaches described in [6], [7] rely on advanced MAC/PHY protocols (such as IEEE 802.11ae) and the resulting algorithms have currently not been implemented. The multicast distribution on wireless mesh networks (WMN) is addressed in [8] and [9], but the suggested solutions rely on proposed advanced features for the data-link layer. [10] presents a

PeerStreamer: assumptions and notation

In this section we introduce and discuss the notation used, the key hypotheses for the theoretic analysis, and some general assumptions on P2P live streaming. We will also review the basic principles of PeerStreamer.

We call the WCN network the underlay to distinguish it from the overlay network built by PeerStreamer. The underlay is supposed to be connected and we assume each node knows the next hop and the distance (the hop count or the weighted hop count depending on the routing protocol) to

Building an optimal overlay

The overlay is built in a distributed way, each Pi choses Ni(t) independently. Building the overlay is one of the major task in a P2P application [14], PeerStreamer is based on the newscast+ [20] protocol that uses low-throughput signalling messages that are exchanged by the peers to share their local knowledge of S. Without entering into details, it is sufficient to say that newscast+ periodically provides each peer with a partial list of all the peers forming the overlay. The choice of the

Video distribution

In live streaming the source is a special peer Ps that provides the media content. In PeerStreamer, Ps concatenates one or more frames (audio and video separately) into chunks and injects them in the overlay. Chunks are numbered; C is the set of all chunks generated by Ps and Ch is the hth chunk. For each chunk Ch there is a delivery deadline dlh related to the playout time, after which Ch is no more useful and it’s “lost” for the application. After its generation at time t, a chunk Ch is

Experimental setup

We performed experiments using two different platforms, the Community-Lab offered by the CONFINE project and a local emulation realized using the Mininet framework. In this section we briefly review both the platforms. The terms node and router are synonymous and identify a network node in the WCN, which is obviously also a router; they are used interchangeably in the following. The term research device (or RD) are instead the general “containers” of the Community-Lab, so the term is used only

Experimental results

We present results for several measurement campaigns, divided between the experiments on the Community-Lab, with the number of peers limited by the research device availability, and without the possibility of cross-layer optimization, and the experiments on Mininet, where these limitations are not present.

Conclusions

Live video distribution is one of the applications that WCNs supports with difficulty, because the interconnection bandwidth toward the Internet to exploit cloud-based systems is often scarce, and because normally there are no resource-rich data centers inside the WCN, which is normally built bottom-up and with a flat architecture.

The use of P2P solutions matches well the characteristics of the WCNs, as they do not require a single node or data center with a lot of resources that distributes to

Luca Baldesi Received his MSc degree in computer engineering from the University of Florence and won a one-year research grant at the Department of Information Engineering and Computer Science at the University of Trento, Italy. He has been mainly working on distributed networking and peer-to-peer communication. He is currently a PhD student at the University of Trento.

References (33)

  • V.C. Borges et al.

    The impact of interference-aware routing metrics on video streaming in wireless mesh networks

    Ad Hoc Netw.

    (2011)
  • L. Maccari et al.

    Betweenness estimation in OLSR-based multi-hop networks for distributed filtering

    J. Comput. Syst. Sci.

    (2014)
  • S. Traverso et al.

    A performance comparison of hose rate controller approaches for P2P-TV applications

    Comput. Netw.

    (2014)
  • L. Maccari et al.

    A week in the life of three large wireless community networks

    Ad Hoc Netw. Part B

    (2015)
  • P. Frangoudis et al.

    Wireless community networks: an alternative approach for nomadic broadband network access

    IEEE Commun. Mag.

    (2011)
  • A. Neumann et al.

    Community-Lab: architecture of a community networking testbed for the future Internet

    Proceedings of IEEE 8th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob 2012), Barcelona, SP, October 2012

    (2012)
  • L. Baldesi et al.

    Live P2P streaming in CommunityLab: experience and insights

    Proceedings of the 13th IEEE/IFIP Mediterranean Ad Hoc Networking Workshop (MedHocNet 2014), June 2014, Piran, SLO

    (2014)
  • L. Baldesi et al.

    Improving P2P streaming in Community-Lab through local strategies

    Proceedings of the 10th IEEE International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob 2014), October 2014, Larnaca, Cyprus

    (2014)
  • B. Lantz et al.

    A network in a laptop: rapid prototyping for software-defined networks

    Proceedings of the 9th ACM SIGCOMM Workshop on Hot Topics in Networks (HOTNETS IX), Oct. 2010, Monterey, Canada

    (2010)
  • Y. Andreopoulos et al.

    Cross-layer optimized video streaming over wireless multihop mesh networks

    IEEE J. Sel. Areas Commun.

    (2006)
  • H.-P. Shiang et al.

    Multi-user video streaming over multi-hop wireless networks: a distributed, cross-layer approach based on priority queuing

    IEEE J. Sel. Areas Commun.

    (2007)
  • R. Campos et al.

    WiFIX+: a multicast solution for 802.11-based wireless mesh networks

    Proceedings of the 8th IEEE/IFIP International Conference on Wireless On-Demand Network Systems and Services (WONS 2011), Jan. 2011, Bardonecchia, Italy

    (2011)
  • A. Russo et al.

    Protocol independent multicast: from wired to wireless networks

    Proceedings of IEEE International Conference on Computing, Networking and Communications (ICNC 2013), January 2013, San Diego, CA, US

    (2013)
  • N. Mastronarde et al.

    Collaborative resource exchanges for peer-to-peer video streaming over wireless mesh networks

    IEEE J. Sel. Areas Commun.

    (2007)
  • S. Mantzouratos et al.

    Survey of cross-layer proposals for video streaming over mobile ad hoc networks (MANETs)

    Proceedings of IEEE International Conference on Telecommunications and Multimedia (TEMU 2012), July 2012, Heraklion, Grece

    (2012)
  • R.J. Lobb et al.

    Adaptive overlay topology for mesh-based P2P-TV systems

    Proceedings of the 18th International Workshop on Network and Operating Systems Support for Digital Audio and Video (NOSSDAV), June 2009, Williamsburg, VA, USA

    (2009)
  • Cited by (15)

    • Enhancing BitTorrent for efficient interactive video-on-demand streaming over MANETs

      2021, Journal of Network and Computer Applications
      Citation Excerpt :

      Rather than that, the rarest (i.e., the least replicated) pieces are firstly requested, hence preventing the continuous file playback. More recently, wireless networks have been playing an increasingly important role to support on-demand streaming applications (e.g., (Hassan and Farooq, 2016; Baldesi et al., 2015; Maccari et al., 2017; Zhang et al., 2019)). In other words, streaming and wireless technologies are interrelated nowadays and will likely continue to be in the upcoming years.

    • Where have all the MPRs gone? On the optimal selection of Multi-Point Relays

      2018, Ad Hoc Networks
      Citation Excerpt :

      The consequence is the possibility to realize city-wide mesh networks made of hundreds of nodes. Some of them have been deeply analyzed in the recent literature [1–3], highlighting advantages, results achieved, but also critical points to be solved [4,5]. The trend to build large wireless metropolitan (and beyond) mesh networks continues, helped by the evolution of 802.11.

    • Optimized P2P streaming for wireless distributed networks

      2017, Pervasive and Mobile Computing
      Citation Excerpt :

      This means that even if the quadratic optimization remains NP, in the analyzed graphs the number of available disjoint paths between two peers is low and the space of the solutions of the optimization problem is small enough for all the optimization strategies to be very close. More results and discussions on this phenomenon can be found [1]. Finally in Fig. 5c we report the mean connectivity of the overlay computed by each strategy.

    View all citing articles on Scopus

    Luca Baldesi Received his MSc degree in computer engineering from the University of Florence and won a one-year research grant at the Department of Information Engineering and Computer Science at the University of Trento, Italy. He has been mainly working on distributed networking and peer-to-peer communication. He is currently a PhD student at the University of Trento.

    Leonardo Maccari is a Post Doc research assistant at the DISI department, he received a degree and a Ph.D in the Faculty of Computer Science Engineering in the University of Florence in 2004 and 2009. He has been awarded with a Marie Curie COFUND grant for the PAF-FPE project for the period 2011–2014. He is an IEEE member and co-authored about 35 publications in refereed conferences, journals and book chapters. He is also among the authors of three patents. The focus of its research is network privacy and security in distributed environments such as wireless mesh/ad-hoc networks and peer-to-peer networks.

    Renato Lo Cigno is an associate professor at the Department of Computer Science and Telecommunications (DISI) of the University of Trento, Italy, where he leads a research group in computer and communication networks. He received a degree in Electronic Engineering from Politecnico di Torino in 1988, where he worked until 2002. From June 1998 to February 1999, he was with the Computer Science Department, University of California, Los Angeles, as a visiting scholar. His current research interests are in performance evaluation of wired and wireless networks, modeling and simulation techniques, congestion control, P2P networks and networked systems in general. Renato Lo Cigno has been an Area Editor for Computer Networks for five years, and has been involved in the organization of several IEEE and ACM Conferences, including Infocom and P2P. He is a senior member of IEEE and member of ACM and has co-authored more than 130 papers all in international, peer reviewed venues.

    This work was partially funded by the European Union’s Seventh Framework Programme for research, technological development and demonstration under Grant agreement no. 288535 “CONFINE” – Open Call 1 project “OSPS”. During the work preparation, Leonardo Maccari has been partially supported by the project “Wireless Community Networks: aspetti giuridici, sociologici e tecnologici di un nuovo fenomeno d’aggregazione sociale” funded by Fondazione CaRiTro.

    View full text