Cost-Effective Multi-Mode Offloading with peer-assisted communications
Introduction
Due to the recent proliferation of smartphones and tablets, mobile data traffic has been explosively growing and pushing the capacity limits of cellular networks In particular, smartphones equipped with high-resolution cameras are sold globally at astounding rates, and content generated by mobile users significantly rises fueled by popular social networking and cloud storage applications such as Facebook, and Dropbox. With the global mobile data traffic forecasted to increase 13 times within the next 5 years [1], cellular network providers need to find cost-effective ways to handle the growing traffic demands with expected Quality of Service (QoS) levels. Infrastructure upgrade is the intuitive solution to deal with the congestion problem; however, it is insufficient as a financially sustainable solution. Offloading mobile data through WiFi networks is the most promising solution to tackle the congestion problem because the cost of WiFi access points (APs) required for the offloading is relatively low, WiFi technology uses unlicensed bands and offloading can satisfactorily serve delay-tolerant types of data traffic. So far, research efforts have been focused on either on-the-spot offloading (OTSO) [2], where data is offloaded only over an immediately available WiFi connection, or delay-tolerant offloading (DTO) [3], where the transmission is delayed for some time in case the user encounters an offloading opportunity later. Both offloading types offer significant benefits, mainly in environments with stable WiFi coverage.
In this paper, we target dense urban environments and propose Cost-Effective Multi-Mode Offloading (CEMMO) that enhances OTSO and DTO with a multi-hop peer-assisted offloading mode (PAO), where the offloaded traffic is delivered through intermediate mobile devices. To the best of our knowledge, PAO is the first peer-assisted offloading solution that is implemented not only for offloading data from the uplink, but also independent of its content or popularity. Given the delay-tolerant nature of a significant portion of the cellular network traffic, an offloading solution that enhances the existing delay-tolerant functionality with peer-assisted offloading (PAO) will leverage the overall offloading capability and, therefore, increase the total amount of data that is offloaded from cellular networks. For PAO, we develop a new data-forwarding scheme that uses limited flooding. By applying a mobility and connectivity prediction model based on a Markov process, CEMMO allows the cellular operator to select the most effective mode for communicating mobile data with different delay constraints based on the estimated costs of delivery through each of the three modes, i.e. cellular delivery, DTO or PAO. OTSO is always exploited when available. The cost reduction leads to lower prices that attract new users, increasing the revenue of the operator. In scenarios with extended WiFi coverage, CEMMO offloads up to 59% of the mobile data traffic and reduces transfer cost per MB up to 16% over DTO. With an energy optimization, CEMMO achieves up to 31% improvement on energy consumption over OTSO.
The novelty of CEMMO lies in the introduction of the first peer-assisted method for offloading traffic from the uplink, regardless of the offloaded content and its popularity, and the selection of the most cost-efficient alternative for data transmission, as defined by each operator. The contributions of the paper can be summarized as follows:
- i.
We design and evaluate CEMMO, a cost-effective scheme for mobile data offloading that integrates multiple modes of operation: cellular delivery, DTO, and PAO.
- ii.
We propose a mobility and connectivity prediction model based on a Markov process and we develop a forwarding scheme for PAO with low storage and energy overhead.
- iii.
CEMMO allows the cellular operators to automatically select the transfer policy (i.e. cellular delivery, DTO or PAO) that provides more gains. The overall cost, as defined by each operator, can include transfer and energy costs, incentives to motivate user participation, etc.
- iv.
CEMMO provides cellular operators with knowledge on the amount of data offloaded by each user. Such knowledge is currently unavailable and will help cellular operators design the expansion of their network accordingly.
The rest of the paper is organized as follows. In Section 2, we describe the proposed mobility and connectivity prediction model. Section 3 details DTO and the proposed PAO transfer policy. CEMMO mechanism is presented in Section 4, along with a sample scenario. Section 5 evaluates the performance of CEMMO through simulations. In Sections 6 Adopting CEMMO, 7 Related work, we discuss several practical issues on the adoption of CEMMO and related work, respectively. Finally, we conclude the paper in Section 8.
Section snippets
Mobility and connectivity prediction model
According to our mobility and connectivity prediction model, a large area is divided into smaller regions with unique identifiers, and a time period (e.g. a day) is split into smaller time intervals (e.g. 10-min intervals). The region size and the time interval duration affect the spatio-temporal accuracy of the prediction model. Defining small regions and time intervals improves the accuracy of the model at the expense of increased computational complexity. Users store their own mobility
Delay-tolerant and peer-assisted offloading transfer policies
In the following subsections, we present the delay-tolerant (DTO) and the peer-assisted offloading (PAO) transfer policies based on the proposed mobility and connectivity prediction model. We assume that each node (i.e. mobile device) is equipped with both IEEE 802.11 and 3G wireless interfaces that can run simultaneously, is capable of generating content for upload and is willing to accept certain delays in applications, such as video upload, without sacrificing too much user satisfaction. We
CEMMO mechanism
The main assumptions of CEMMO are that all mobile nodes apply the proposed mobility and connectivity prediction model, and that the cellular operator collects data transfer requests from users, performs online predictions on user mobility and connectivity and decides on the transfer policy that minimizes cost. We assume that users perform OTSO when they have direct access to a WiFi network and are willing to accept a delay in the delivery of some non-urgent data, when proper incentives are
Evaluation
In this section, we provide an extensive evaluation of the Cost-Effective Multi-Mode Offloading (CEMMO) mechanism in comparison to pure on-the-spot offloading (OTSO) and pure delay-tolerant offloading (DTO), which are the state-of-the-art approaches in mobile data offloading. All DTO solutions in literature share the same functionality; only the delay-tolerance threshold changes. Our experiments have been conducted using the Opportunistic Network Environment (ONE) simulator [8], which has been
Adopting CEMMO
In this section, we discuss some issues related to the adoption of CEMMO. First, there exists a need for a mechanism that allows users to set the desired delay tolerance threshold for each application according to its requirements. This can be implemented by exposing a simple application programming interface (API) per application, similar to e-mail refresh rate settings in mobile devices. Alternatively, CEMMO can use application port information to infer a predefined delay tolerance to
Related work
The concept of data offloading through femtocells was the first approach to alleviate the cellular burden by using an alternative access network, however it contends for limited licensed spectrum with cellular networks [13]. The recent advancements in the APIs of mobile operating systems enable seamless transition from 3G to WiFi networks transparently to the user [14], [15], have led to the adoption of OTSO [2], [16] due to its financial and energy efficiency. Solutions that utilize multiple
Conclusions
Cellular networks will not be able to handle the upcoming explosion in mobile data growth mainly due to economic reasons that pose barriers to wide infrastructure upgrades. As an alternative solution, cellular operators seek to offload delay-tolerant traffic from the cellular network utilizing low-cost WiFi connections. We argue that the total amount of offloaded traffic can significantly increase through peer-assisted offloading; intermediate nodes act as relays between the source node and the
Acknowledgments
The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013, FP7-REGPOT-2010-1, SP4 Capacities, Coordination and Support Actions) under Grant Agreement n° 264226 (Project title: Space Internetworking Center – SPICE), the European Commission (FP7-ICT 288021, EINS) and Spanish Ministry of Science and Innovation (RYC-2009-04660).
Ioannis Komnios graduated and received his MSc degree on Computer Networks from the Department of Electrical and Computer Engineering, Democritus University of Thrace, in 2007 and 2009, respectively. Currently he is a PhD candidate at the same department and a researcher at Space Internetworking Center in Xanthi, Greece. Ioannis has a long experience in internetworking in delay/disruptive tolerant environments and routing protocols. His current research interests lie in the area of data
References (35)
SWISH: secure WiFi sharing
Elsevier Comput. Netw.
(2011)Push-and-track: saving infrastructure bandwidth through opportunistic forwarding
Pervas. Mobile Comput.
(2012)- Cisco Visual Networking Index, Global Mobile Data Traffic Forecast Update,...
- et al.
Mobile data offloading: how much can WiFi deliver?
IEEE Trans. Netw.
(2013) - A. Balasubramanian, R. Mahajan, A. Venkataramani, Augmenting mobile 3G using WiFi, in: ACM MobiSys,...
- Android Geofences....
- V. Misra, S. Ioannidis, A. Chaintreau, L. Massouli, Incentivizing peer-assisted services: a fluid Shapley value...
- J. Lee, Y. Yi, S. Chong, Y. Jin, Economics of WiFi offloading: trading delay for cellular capacity, in: IEEE INFOCOM,...
- et al.
An incentive framework for cellular traffic offloading
IEEE Trans. Mobile Comput.
(2014) - A. Keränen, J. Ott, T. Kärkkäinen, The ONE simulator for DTN protocol evaluation, in: SIMUTools’09, Rome,...
Cited by (22)
Weighted proportional fairness and pricing based resource allocation for uplink offloading using IP flow mobility
2016, Ad Hoc NetworksCitation Excerpt :Another example is the division of a UE’s data flow into two sub-flows and the service of each sub-flow by different radio access technologies, as proposed in [8]. In [9], the authors introduce the first peer-assisted method for offloading traffic from the uplink, and the selection of the most cost-efficient alternative access technology for data transmission. A question that arises from the IFOM uplink offloading scheme is how the UEs will offload part of their data through WiFi with fairness, where their different upload data needs and their LTE connection with the eNodeB will be considered, and how the rest of the data will be uploaded through LTE.
Integrated Privacy Preserving Healthcare System Using Posture-Based Classifier in Cloud
2023, Intelligent Automation and Soft ComputingAdaptive cost and energy aware secure peer-to-peer computational offloading in the edge-cloud enabled healthcare system
2021, Peer-to-Peer Networking and ApplicationsTowards Realistic DTN Simulations for Public Transport Networks
2021, 2021 International Conference on COMmunication Systems and NETworkS, COMSNETS 2021Cooperative offloading in D2D-enabled three-tier MEC networks for IoT
2021, Wireless Communications and Mobile Computing
Ioannis Komnios graduated and received his MSc degree on Computer Networks from the Department of Electrical and Computer Engineering, Democritus University of Thrace, in 2007 and 2009, respectively. Currently he is a PhD candidate at the same department and a researcher at Space Internetworking Center in Xanthi, Greece. Ioannis has a long experience in internetworking in delay/disruptive tolerant environments and routing protocols. His current research interests lie in the area of data offloading and less-than-best-effort services.
Fani Tsapeli received a diploma in Electronic and Computer Engineering from the Technical University of Crete, Greece, in 2009 and an MSc in Computer Networks from the Department of Electrical and Computer Engineering, Democritus University of Thrace, Greece, in 2012 under the advisory of Prof. Vassilis Tsaoussidis. From 2010 to 2013, Fani was working as a researcher at the Space Internetworking Center (http://www.spice-center.org/).
Sergey Gorinsky received an Engineer degree from Moscow Institute of Electronic Technology, Zelenograd, Russia in 1994 and M.S. and Ph.D. degrees from the University of Texas at Austin, USA in 1999 and 2003 respectively. From 2003 to 2009, he served on the tenure-track faculty at Washington University in St. Louis, USA. Dr. Gorinsky currently works as a tenured Research Associate Professor at IMDEA Networks Institute, Madrid, Spain. The areas of his primary research interests are computer networking, distributed systems, and network economics. His research contributions include multicast congestion control resilient to receiver misbehavior, analysis of binary adjustment algorithms, efficient fair transfer of bulk data, network service differentiation based on performance incentives, and economic perspectives on Internet interconnections and routing. His work appeared at top conferences and journals such as ACM SIGCOMM, IEEE INFOCOM, ACM CoNEXT, IEEE/ACM Transactions on Networking, and IEEE Journal on Selected Areas in Communications. Sergey Gorinsky delivered keynote addresses at NPSec 2013 and RAIT 2012. He has served on the TPCs (technical program committees) of SIGCOMM (2012), INFOCOM (2006–2014), ICNP (2008, 2010–2014), and other networking conferences. Prof. Gorinsky co-chaired the TPCs of COMSNETS 2013, NetSciCom 2014, E6 2012, HSN 2008, FIAP 2008, served as a TPC vice-chair for ICCCN 2009 and TPC area chair for ICNP 2013.