Elsevier

Computer Networks

Volume 55, Issue 2, 1 February 2011, Pages 399-414
Computer Networks

On Demand Connectivity Sharing: Queuing management and load balancing for User-Provided Networks

https://doi.org/10.1016/j.comnet.2010.08.015Get rights and content

Abstract

We introduce the concept of “On Demand Connectivity Sharing”, which we build on top of User-Provided Networks (UPNs). UPNs were recently proposed as a new connectivity paradigm, according to which home-users share their broadband Internet connection with roaming guests. We enhance this paradigm with incentives, rules and policies, based on which: (i) home-users provide on-demand connectivity only (i.e., they do not explicitly allocate a portion of their bandwidth) and (ii) guest-users utilize resources that remain unexploited from the respective home-users.

We realize the “On Demand Connectivity Sharing” concept through (i) a queuing algorithm that classifies traffic according to its source (i.e., home- or guest-traffic) and prioritizes home- against guest-traffic accordingly and (ii) a probabilistic load-balancing algorithm that guarantees smooth cooperation between home- and guest-users. We show both analytically and through extensive performance evaluation that it is indeed possible for a home-user to share his connection with guest-, roaming-users without any practical impact on his own network performance.

The concept of “On Demand Connectivity Sharing” through User-Provided Networks is expected to receive a lot of attention in the years to come, since it enables a new notion of autonomous and self-organized mobile computing. For example, we gather information regarding the location and range of real WiFi access points in the city center of London and we show that a walking user can receive acceptable services, when acting as a guest-user and gets resources from near-by home-networks.

Introduction

One of the main features of the Future Internet is that it is going to be everywhere, at anytime and for everyone. Two main approaches have been identified till now to achieve that goal: (i) mobile Internet through 3G links (e.g., [1], [2]) and (ii) mobile Internet as an extension of the network itself (e.g., [3], [4]). The first approach assumes that users use telecommunication channels to reach the Internet (i.e., 3G links), while the second assumes extension of the network infrastructure and cooperation among users (possibly in an ad hoc manner) in order to bring connectivity further away from the strict boundaries of the traditional Internet (e.g., VANETs [3], or mesh networks [4]). Mobile operators, however, initially designed and setup their networks to carry voice traffic only. Then SMS text messages came into play and formed a major source of revenue for telecommunication vendors (i.e., very few bytes for a relatively expensive price – almost infinite Return Of Investment) and lately Internet connectivity is provided as part of the user’s contract with the mobile operator. It is questionable though whether the telecommunication network will be able to handle large volumes of data instead of simple voice transfers and short text messages, once billions of users make use of such services. That said, extension of the network itself in order to offload 3G traffic to WiFi networks and achieve ubiquitous connectivity seems to be a more realistic, elegant and scalable approach.1

The first step on that direction has been made with mobile ad hoc and mesh networks (e.g., [4], [5], [6], [7]), where nodes act as relays for messages they are not interested in. These nodes, obviously, consume resources (e.g., energy) to receive and transmit messages further in order to reach the destination, without having explicit gain themselves. In our opinion, this formed the fundamental first step towards the “On Demand Resource Sharing” era. With the evolution of “peer-to-peer” networks [8], users provide access to their local resources (i.e., hard-drive) to other users in order to speed up bulk data transfers. Data transfers take place in a distributed manner keeping congestion away from the main backbone links. Lately, a number of new concepts have attracted the attention of the research community, again in the direction of On Demand Resource Sharing. For example, in Delay-Tolerant Networks (DTNs) [9], [10], nodes store, carry and forward messages in order to deal with large delays, or disrupted connections. Hence, they act as relays and consume resources to forward data they are not particularly interested in, or Online Social Networks [11], [12], where users receive, store, carry and forward messages to and from other users regarding social interests [13] and habits [14], [15].

In this study, we propose that “On Demand Resource Sharing” has to be complemented with “On Demand Connectivity Sharing” in order to: (i) extend the traditional Internet’s connectivity boundaries, (ii) give rise to the Future Mobile Internet and (iii) enable new types of technologies and applications, such as DTNs, CCNs and Online Social Networks. To reach the point where users can share connectivity resources, however, there are a number of issues that need to be investigated, such as for example, incentives for the home-user to offer resources and performance implications due to sharing [16]. We attempt to address the above issues building on the concept of “User-Provided Networks” [17].

In User-Provided Networks (UPNs), a home-user, owner of a broadband Internet connection, provides connectivity to unknown, mobile users that roam within the home-user’s Access Point (AP) connectivity range. In that sense, the home-user is a consumer of Internet resources provided by the Internet Service Provider (ISP) and at the same time provider of Internet services to roaming users. The home-user is, thus, a micro-provider [17]. Throughout this paper, we use the terms “home-user” and “micro-provider” interchangeably to denote the owner of the broadband Internet connection.

According to the authors of [17], the UPN connectivity paradigm needs to: (i) be easily deployable in terms of software and hardware modifications, (ii) provide the appropriate incentives to users to join the community (i.e., the more the micro-providers, the more dense connectivity is), (iii) guarantee that only liable users can access the micro-provider’s resources and moreover, that they do not misbehave and (iv) manage roaming users connectivity-wise in a self-adaptive and autonomous way.

In this study, we show that it is possible for the home-user to share his broadband connection with mobile, roaming users without any noticeable impact on his own performance. In turn, the home-user gains unlimited connectivity (by other micro-providers) when he is out of home or office. We achieve the above by introducing the following two algorithms for UPNs:

  • (1)

    An active queue management algorithm, that classifies and schedules home- and guest-packets according to different priorities. These priorities are designed so that (i) the home-packets are always favored against guest-packets and (ii) guest-packets go through only if there is sufficient space for them at the UPN-AP’s queue. We show that this packet classification and scheduling algorithm guarantees that sharing has practically no impact on the home-user’s performance.

  • (2)

    A simple load-balancing mechanism that routes guest-traffic through the least-congested UPN-AP. That is, if a roaming user is within range of more than one UPN-APs, then his traffic is routed through the least-utilized AP. This algorithm is also complemented by a probabilistic switching scheme to guarantee that not all mobile-users switch to the same UPN-AP.

Therefore, the incentive for the home-user to share his broadband Internet connection is simple: he shares a portion of his bandwidth seamlessly, when at home and gains unlimited connectivity from other home-users, when mobile. We call this scheme Offer Nothing – Gain Something. Of course, this “Something” depends on the density of the available access points as well as on the bandwidth utilization of the respective home-users. That is, connectivity may be poor and intermittent, making DTN technology an essential complementary part for UPNs. The argument here, however, is that even poor, opportunistic connectivity is enough to download a web page, an e-mail or a map, which are popular mobile applications; we verify this claim by realistic simulation experiments. We argue that connectivity is for mobile computing what bandwidth is for the wired core of the Internet. Therefore, instead of high-speed core Internet links, the backbone for a mobile environment is the area of dense connectivity. We contend that On Demand Connectivity Sharing, through User-Provided Networks, comes as the natural evolution of On Demand Resource Sharing schemes; these new connectivity and data management paradigms that enable and support the notion of autonomous and self-organizing networks are essential for the realization of the Future Internet that inherently supports mobility.

We clarify the following:

  • A mobile user that connects to different access points and stays there connected for some time (e.g., from the hotel-AP to the restaurant-AP and later on to the conference venue-AP) requires manual network discovery and re-connection; this notion of mobility has already been introduced by FON [18], OpenSpark [19] and Whisher [20]. However, users may be mobile 100% of the time, e.g., a walking-user roaming on the street or inside a vehicle. This user needs autonomous (i.e., self-adaptive) and on-the-fly connectivity to several different APs. Here, we focus on the second case of mobility, which we consider as an enabler for the future mobile Internet.

  • Service differentiation through active queue management has received a lot of attention in the past few years (e.g., [21], [22], [23]), mainly in order to prioritize high-paying customers’ traffic against low-paying ones’, or to boost the performance of non-elastic, or non-congestive applications against bursty TCP transfers. Load-balancing, on the other hand, has been studied mainly in the context of server farms (e.g. [24]), peer-to-peer networks (e.g., [25]) and multipath routing mechanisms (e.g., [26], [27], [28]). The novelty introduced herein is the actual marriage and application of these techniques in the context of UPNs in order to realize new communication technologies and connectivity paradigms that explicitly support mobile computing. To the best of our knowledge, this is the first study that puts these two research fields under a different context, that of connectivity sharing and elaborates on their potential to set the foundations for the realization of the “On Demand Connectivity Sharing” through User-Provided Networks.

The rest of the paper is organized as follows: we begin with a description of the incentives that would convince the end-user to switch to the UPN scheme and share his broadband Internet connection with mobile users (Section 2). In Section 3, we present our design proposals; namely the UPN Queuing (UPNQ) algorithm for service differentiation between the home- and the guest-user and the UPN Load Balancing (UPNLB) algorithm for load-balancing among roaming users. Next, in Section 4, we describe our experimental setup and present our simulation results. Finally, in Section 5, we discuss open issues and future challenges and conclude the paper in Section 6.

Section snippets

The Offer Nothing – Gain Something Sharing scheme

We consider that the success of the “On Demand Connectivity Sharing” scheme depends mainly on the specific incentives given to home-users in order to motivate them to shift into this new connectivity paradigm. This process consists of two main steps: (i) offer attractive “deals” to prospective users and (ii) design the corresponding algorithms to guarantee smooth and scalable operation of the proposed offers.

We attempt to pose and answer questions that would naturally come to a user’s mind when

Service differentiation for UPNs

When joining the UPN community, each user is provided with a username and a password. This pair of identification is used to access the Internet: if the user connects to his own access point, then he is identified as the home-user or the micro-provider, while if he is mobile and connects to an unknown access point, he is identified as the guest-user. In turn, whatever packets are sent or received by that user are classified accordingly. This way, we implement a packet-classification algorithm

Simulation results

For the first two simulation scenaria included herein (i.e., Sections 4.1 Scenario 1: Home-user traffic with “On/Off” periods, 4.2 Scenario 2: Gradually decreasing home-user traffic), we begin with the simplistic topology depicted in Fig. 1. The purpose is to investigate and understand the performance of the proposed framework in a simple topology. Next, for the third and the fourth scenario, we use more realistic topologies; the specifics of these last two topologies are given in Sections 4.3

Open issues

In this section, we discuss open issues related to our proposed UPN Protocols and their potential extensions. The target is to identify important future challenges for the “On Demand Connectivity Sharing” framework proposed previously.

Conclusions

We have introduced the concept of “On Demand Connectivity Sharing”, according to which home-users can share their connection seamlessly with guest-, roaming-users. We have built our framework on top of the recently proposed User-Provided Networking [17] connectivity paradigm. We proved and showed that both in theory and in practice careful design can guarantee seamless resource sharing for the home-user, while the guest can still enjoy acceptable performance. Our proposed UPN Queuing (UPNQ) and

Ioannis Psaras received a diploma in Electrical and Computer Engineering from Democritus University of Thrace, Greece in 2004, and the Ph.D. degree from the same institute in 2008. He won the Ericsson Award of Excellence in Telecommunications for his diploma dissertation in 2004. Ioannis has worked, as a research intern at DoCoMo Eurolabs (May–September 2005) and at Ericsson Eurolab (May–September 2006). His research interests are in the areas of Congestion/Flow Control, Transport-layer

References (42)

  • I.F. Akyildiz et al.

    Wireless mesh networks: a survey

    Comput. Networks

    (2005)
  • I.F. Akyildiz et al.

    Crahns: cognitive radio ad hoc networks

    Ad Hoc Networks

    (2009)
  • T. Halonen et al.

    GSM, GPRS and EDGE Performance: Evolution towards 3G/UMTS

    (2003)
  • H. Holma et al.

    Wcdma for Umts

    (2000)
  • J. Bernsen et al.

    Routing Protocols for Vehicular Ad Hoc Networks That Ensure Quality of Service

  • S. Corson, J. Macker, Mobile Ad hoc Networking (MANET): Routing Protocol Performance Issues and Evaluation...
  • I. Akyildiz et al.

    A survey on wireless mesh networks

    IEEE Commun. Mag.

    (2005)
  • S. Androutsellis-Theotokis et al.

    A survey of peer-to-peer content distribution technologies

    ACM Comput. Surv.

    (2004)
  • K. Fall

    A delay-tolerant network architecture for challenged internets

  • I. Psaras, L. Wood, R. Tafazolli, Delay-/Disruption-Tolerant Networks: State of the Art and Research Challenges,...
  • J. Scott, P. Hui, J. Crowcroft, C. Diot, Haggle: A networking architecture designed around mobile users, in:...
  • P. Hui et al.

    Pocket-switched networks and human mobility in conference environments

  • P. Hui et al.

    Bubble Rap: social-based forwarding in delay-tolerant networks

  • A.J. Mashhadi, S.B. Mokhtar, L. Capra, Habit: leveraging human mobility and social network for efficient content...
  • L. McNamara et al.

    Media sharing based on colocation prediction in urban transport

  • L. Mamatas, I. Psaras, Incentives and algorithms for User-Provided Networks, in: Proceedings of the ACM SIGCOMM...
  • R. Sofia et al.

    User-provided networks: consumer as provider

    IEEE Commun. Mag.

    (2008)
  • FON Community,...
  • OpenSpark Community,...
  • Whisher Community,...
  • I. Stoica, S. Shenker, H. Zhangm, Core-stateless fair queueing: achieving approximately fair bandwidth allocations in...
  • Cited by (16)

    • Energy efficient mobile video streaming using mobility

      2016, Computer Networks
      Citation Excerpt :

      This is to say, that only few media fragments are delivered during bad channel conditions while larger volumes exchanged only during good conditions. It should be emphasized here that the proposed scheme differs from drive-through Internet [7,8]. The latter concept deals with content delivery when there is intermittent connectivity while here we assume that connectivity is perpetual.

    • An expert system for real-time traffic management in wireless local area networks

      2014, Expert Systems with Applications
      Citation Excerpt :

      Common approaches to network selection are based on estimated QoS parameters for each available network. Adamopoulou, Demestichas, Koutsorodi, and Theologou (2005), Bari and Leung (2007), Psaras and Mamatas (2011), Song and Jamalipour (2005), Wang, Katz, and Giese (1999), Wilson, Lenaghan, and Malyan (2005), Xing and Venkatasubramanian (2005) and Yahiya and Chaouchi (2009) consider network QoS parameters such as delay and capacity. In the literature, researchers have also proposed more user-centric criteria, among them the power consumption of devices (see Yahiya & Chaouchi (2009) and Petander (2009)) or the cost of network use (see Adamopoulou et al. (2005), Bari & Leung (2007), Wilson et al. (2005) and Alkhawlani & Ayesh (2008)).

    • Performance evaluation of two-link multirate loss models with restricted accessibility

      2020, Journal of Telecommunications and Information Technology
    • Call blocking probabilities in a two-link loss model with restricted accessibility

      2019, ConTEL 2019 - 15th International Conference on Telecommunications, Proceedings
    View all citing articles on Scopus

    Ioannis Psaras received a diploma in Electrical and Computer Engineering from Democritus University of Thrace, Greece in 2004, and the Ph.D. degree from the same institute in 2008. He won the Ericsson Award of Excellence in Telecommunications for his diploma dissertation in 2004. Ioannis has worked, as a research intern at DoCoMo Eurolabs (May–September 2005) and at Ericsson Eurolab (May–September 2006). His research interests are in the areas of Congestion/Flow Control, Transport-layer Protocols, Space and Deep-Space Communications, Delay-/Disruption-Tolerant Networks (DTNs), User-Provided and User-Centric Networks and Content-Centric Networks (CCNs). He is currently working as a post-doctoral researcher at the Networks and Services Research Lab (NSRL) at the Electronic and Electrical Engineering Department of University College London (UCL). Before, he was working as a researcher at the Center for Communications and Systems Research (CCSR) of the University of Surrey (2008–2010). Ioannis participates in the Technical Program Committees for several networking conferences, such as IFIP Networking 2009, IEEE LCN 2010, IEEE ICCCN 2009, 2010, IEEE Globecom 2009, 2010 and several others. He also serves as a reviewer for several journals, such as Elsevier Computer Networks, IEEE Transactions for Mobile Computing, IEEE Communications Magazine, among others. Further information can be found at: http://www.ee.ucl.ac.uk/∼uceeips/.

    Lefteris Mamatas is an Adjunct Lecturer at University of Macedonia. Before that, he was a Research Associate at University College London (UCL). He received his Ph.D. in June 2008 from Democritus University of Thrace. He has a long experience in protocol engineering (e.g., energy efficient transport/network layer protocols, protocols for DTN & User-Provided Networks etc), experimental & theoretical analysis on performance evaluation and test-bed implementation. He was a founder of Araneous Internet Services (an IT company situated in North Greece). In 2009, he chaired the IEEE Workshop on the Emergence of Delay-/Disruption-Tolerant Networks (E-DTN). He published 10 journal papers, 17 conference papers and 1 book. More information can be found at: http://users.uom.gr/∼emamatas.

    View full text