Name
Abstract | ||
|---|---|---|
Traffic monitoring is instrumental to a number of applications such as traffic engineering, QoS routing, anomaly detection and so on. With an accurate global view, software defined networking (SDN) has the capability to offer flexible, non-intrusive flow measurement by using wildcard matching in both direct per-flow and indirect aggregated manners. As a result, the complete and fine-grained traffic matrix (TM) monitoring, which is a challenge in traditional large-scale sensor networks, becomes more accessible. However, exiting SDN monitoring solutions have a poor trade-off between the resource-hungry nature of full sampling and limited accuracy of TM inference. Thus, in this paper, we aim to address this issue by developing CFlow, a lightweight compressive flow statistics collection (FSC) scheme for SDNs. By taking advantage of the low-rank and short-term stability features of real-world TMs, CFlow selectively samples flow statistics through custom-tailored wildcard rules, with which the final TMs are recovered via the matrix completion technique. Moreover, since the accurate measurement of large flows can improve the overall TM estimation performance, CFlow successively learns these informative flows with additional observation from both per-flow statistics collection and the latest recovered TMs. Simulation results based on real TMs demonstrate that CFlow can not only provide fine-grain visibility into network traffic but also avoid considerable monitoring overhead. |
Year | DOI | Venue |
|---|---|---|
2019 | 10.1109/ICC.2019.8761224 | IEEE International Conference on Communications |
Field | DocType | ISSN |
Anomaly detection,Visibility,Wildcard,Matrix completion,Computer science,Quality of service,Real-time computing,Software-defined networking,Wireless sensor network,Traffic engineering | Conference | 1550-3607 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Mingyan Li | 1 | 13 | 5.81 |
| Cai-Lian Chen | 2 | 831 | 98.98 |
| Cunqing Hua | 3 | 273 | 29.95 |
| Xinping Guan | 4 | 2791 | 253.38 |