The evaluation demonstrates the feasibility and practicality of our prototype imple- mentation in WiFi networks. Our system works well for static clients and makes accurate decisions based on some context information. However, the test results also show some potential problems, and we discuss them in this section.
Offloading Mobility Prediction
The experiment data on client overhead show that our current mobility prediction algorithm and implementation may not work for some new devices. For example, it takes about 13 seconds to finish 3-time WiFi scanning on Samsung Galaxy S5, which is too long and makes offloading impractical. In addition, it treats all different users as the same type, and requests all of them to perform multi-time scanning. This may be also unnecessary and inefficient. In one word, we may need to design a new mechanism to quickly give short-term movement prediction on all kinds of devices when it is necessary.
Cellular Testing
In our current implementation, cellular access points are simplified and abstracted as the same wireless resources as WiFi access points, and we only use WiFi networks to show our system’s feasibility. Although we try to design a general system for different networks, further tests on an extended platform both with WiFi and cellular access points are required. Cellular and WiFi are still different techniques, and there may be some limitations and requirements only appearing in cellular networks.
6
Conclusion and Future Work
In this paper, we consider the traffic offloading problem and propose an SDN-based system which performs offloading by utilizing different kinds of wireless context information. The system design is motivated by a real-world measurement study on wireless network accesses. To make beneficial offloading decisions not only for the system, but also for clients, our system collects information from two sides and evaluates them in a comprehensive way. The feasibility and practicality of our design system are verified through a prototype implementation, and we demonstrate its performance and overhead with real-world test cases. The evaluation results show that our proposal system may achieve optimal offloading by considering various factors, and avoid "naive" offloading.
In addition to context-based offloading decision making, we explore the idea of managing networks in a software-based approach, and develop our system with SDN techniques. By expressing different functions as software applications, we demonstrate that it is easy to introduce new ideas and turn them into reality in an SDN system like our platform.
As the first step to the software-based traffic offloading, there are still some problems, as well as possibilities in our current system. We leave them as future work, and hope to solve them later:
• Mobility Prediction: our current offloading mobility prediction brings so long latencies on some new Android devices that additional support is required. Furthermore, we treat all devices as the same type and perform identical 3- time scanning, which may be inefficient and inaccurate on some devices. For example, the 3-time scanning costs extra energy but is probably unnecessary for static clients. Users with vehicular mobility shall avoid WiFi switching because it brings more limited coverage. We shall not only modify current prediction algorithm, but distinguish different types of users in the future work.
• Cellular Network Evaluation: In our current implementation, no cellular access point is really deployed and tested. Though we emphasize that our design system is general and shall support different network types, further evaluation and validation are still necessary. Porting our current local agents to cellular base stations may also raise some new requirements.
• Bandwidth Prediction: As one of the key features in our platform, the con- troller is able to make offloading decisions based on AP bandwidth information. However, we only collect real-time bandwidth data before user switches net- works, which may be not enough. For example, if a client is first moved to an idle AP in one room, then a crowd of users rush in and connect to this AP, the first client may still suffer from lower bandwidth. To avoid this problem, we may need to perform short-term bandwidth and traffic prediction for each AP. Currently we are working on this issue, and trying to explore whether we can make more accurate offloading decisions with a history-based traffic prediction.
• Scalability: the current prototype implementation only involves 2 controllable APs and one OpenFlow switch. If the size of a network continues to increase, system overhead and resource costs for the central controller to manage the network can be expensive. The scalability shall be explored in our future work. • Other Applications: there are many possible extensions for our platform, e.g. traffic monitoring and access control. As one of our future tasks, we plan to explore practical mechanisms for dynamically channel adjustment in our platform.
References
IEEE09 Ieee standard for information technology– local and metropolitan area networks– specific requirements– part 11: Wireless lan medium access control (mac) and physical layer (phy) specifications amendment 5: En- hancements for higher throughput. IEEE Std 802.11n-2009 (Amend- ment to IEEE Std 802.11-2007 as amended by IEEE Std 802.11k- 2008, IEEE Std 802.11r-2008, IEEE Std 802.11y-2008, and IEEE Std 802.11w-2009), pages 1–565.
BC03 Balazinska, M. and Castro, P., Characterizing mobility and network usage in a corporate wireless local-area network. Proceedings of the 1st International Conference on Mobile Systems, Applications and Services, MobiSys ’03, New York, NY, USA, 2003, ACM, pages 303–316.
BMV10 Balasubramanian, A., Mahajan, R. and Venkataramani, A., Augment- ing mobile 3g using wifi. Proceedings of the 8th International Conference on Mobile Systems, Applications, and Services, MobiSys ’10, New York, NY, USA, 2010, ACM, pages 209–222.
CFP+07 Casado, M., Freedman, M. J., Pettit, J., Luo, J., McKeown, N. and Shenker, S., Ethane: Taking control of the enterprise. SIGCOMM Com- put. Commun. Rev., Vol. 37, No. 4(2007), pages 1–12.
Cisco14 Cisco visual networking index: Global mobile data traffic forecast up- date, 2013–2018, 5 February 2014
DHM+13 Dixit, A., Hao, F., Mukherjee, S., Lakshman, T. and Kompella, R., To-
wards an elastic distributed sdn controller. Proceedings of the Second ACM SIGCOMM Workshop on Hot Topics in Software Defined Net- working, HotSDN ’13, New York, NY, USA, 2013, ACM, pages 7–12. DHX+13 Ding, A. Y., Han, B., Xiao, Y., Hui, P., Srinivasan, A., Kojo, M. and
Tarkoma, S., Enabling energy-aware collaborative mobile data offload- ing for smartphones. 10th Annual IEEE Communications Society Con- ference on Sensor, Mesh and Ad Hoc Communications and Networks, SECON ’13, New Orleans, LA, USA, June 2013, pages 487–495. Eri13 Erickson, D., The beacon openflow controller. Proceedings of the Sec-
Networking, HotSDN ’13, New York, NY, USA, 2013, ACM, pages 13– 18.
GHM+05 Greenberg, A., Hjalmtysson, G., Maltz, D. A., Myers, A., Rexford, J., Xie, G., Yan, H., Zhan, J. and Zhang, H., A clean slate 4d approach to network control and management. SIGCOMM Comput. Commun. Rev., Vol. 35, No. 5(2005), pages 41–54.
GKP+08 Gude, N., Koponen, T., Pettit, J., Pfaff, B., Casado, M., McKeown,
N. and Shenker, S., NOX: Towards an operating system for networks. SIGCOMM Comput. Commun. Rev., Vol. 38, No. 3(2008), pages 105– 110.
GPLK13 Gudipati, A., Perry, D., Li, L. E. and Katti, S., Softran: Software de- fined radio access network. Proceedings of the Second ACM SIGCOMM Workshop on Hot Topics in Software Defined Networking, HotSDN ’13, New York, NY, USA, 2013, ACM, pages 25–30.
HHK+12 Han, B., Hui, P., Kumar, V. S. A., Marathe, M. V., Shao, J. and
Srinivasan, A., Mobile data offloading through opportunistic communi- cations and social participation. IEEE Transactions on Mobile Com- puting, Vol. 11, No. 5(2012), pages 821–834.
HRA+10 Higgins, B. D., Reda, A., Alperovich, T., Flinn, J., Giuli, T. J., Noble, B. and Watson, D., Intentional networking: Opportunistic exploitation of mobile network diversity. Proceedings of the Sixteenth Annual Inter- national Conference on Mobile Computing and Networking, MobiCom ’10, New York, NY, USA, 2010, ACM, pages 73–84.
HYG12 Hassas Yeganeh, S. and Ganjali, Y., Kandoo: A framework for efficient and scalable offloading of control applications. Proceedings of the First Workshop on Hot Topics in Software Defined Networks, HotSDN ’12, New York, NY, USA, 2012, ACM, pages 19–24.
JLVR13 Jin, X., Li, L. E., Vanbever, L. and Rexford, J., Softcell: Scalable and flexible cellular core network architecture. Proceedings of the Ninth ACM Conference on Emerging Networking Experiments and Technolo- gies, CoNEXT ’13, New York, NY, USA, 2013, ACM, pages 163–174.
KCG+10 Koponen, T., Casado, M., Gude, N., Stribling, J., Poutievski, L., Zhu, M., Ramanathan, R., Iwata, Y., Inoue, H., Hama, T. and Shenker, S., Onix: A distributed control platform for large-scale production net- works. Proceedings of the 9th USENIX Conference on Operating Sys- tems Design and Implementation, OSDI’10, Berkeley, CA, USA, 2010, USENIX Association, pages 1–6.
KMC+00 Kohler, E., Morris, R., Chen, B., Jannotti, J. and Kaashoek, M. F.,
The click modular router. ACM Trans. Comput. Syst., Vol. 18, No. 3(2000), pages 263–297.
LLY+10 Lee, K., Lee, J., Yi, Y., Rhee, I. and Chong, S., Mobile data offload- ing: How much can wifi deliver? Proceedings of the 6th International Conference, Co-NEXT ’10, New York, NY, USA, 2010, ACM, pages 26:1–26:12.
MAB+08 McKeown, N., Anderson, T., Balakrishnan, H., Parulkar, G., Peterson,
L., Rexford, J., Shenker, S. and Turner, J., Openflow: Enabling inno- vation in campus networks. SIGCOMM Comput. Commun. Rev., Vol. 38, No. 2(2008), pages 69–74.
MPC+08 Murty, R., Padhye, J., Chandra, R., Wolman, A. and Zill, B., Design- ing high performance enterprise wi-fi networks. Proceedings of the 5th USENIX Symposium on Networked Systems Design and Implementa- tion, NSDI’08, Berkeley, CA, USA, 2008, USENIX Association, pages 73–88.
MPWW10 Murty, R., Padhye, J., Wolman, A. and Welsh, M., Dyson: An architec- ture for extensible wireless lans. Proceedings of the 2010 USENIX Con- ference on USENIX Annual Technical Conference, USENIXATC’10, Berkeley, CA, USA, 2010, USENIX Association.
NNH+14 Nirjon, S., Nicoara, A., Hsu, C.-H., Singh, J. P. and Stankovic, J. A., Multinets: A system for real-time switching between multiple network interfaces on mobile devices. ACM Trans. Embed. Comput. Syst., Vol. 13, No. 4s(2014), pages 121:1–121:25.
OF09 Openflow switch specification – version 1.0.0, 31 December 2009 OF13 Openflow switch specification – version 1.4.0, 14 October 2013
SSZM+12 Suresh, L., Schulz-Zander, J., Merz, R., Feldmann, A. and Vazao, T., Towards programmable enterprise wlans with odin. Proceedings of the First Workshop on Hot Topics in Software Defined Networks, HotSDN ’12, New York, NY, USA, 2012, ACM, pages 115–120.
TG10 Tootoonchian, A. and Ganjali, Y., Hyperflow: A distributed con- trol plane for openflow. Proceedings of the 2010 Internet Network Management Conference on Research on Enterprise Networking, IN- M/WREN’10, Berkeley, CA, USA, 2010, USENIX Association.
YSK+10 Yap, K.-K., Sherwood, R., Kobayashi, M., Huang, T.-Y., Chan, M., Handigol, N., McKeown, N. and Parulkar, G., Blueprint for introducing innovation into wireless mobile networks. Proceedings of the Second ACM SIGCOMM Workshop on Virtualized Infrastructure Systems and Architectures, VISA ’10, New York, NY, USA, 2010, ACM, pages 25–32.