Side-Channel Attacks in SDNs: What the Literature Reveals

Software-Defined Networking decouples the control plane from the data plane, enabling flexible, programmable network management. That architectural separation, however, creates new side-channel attack surfaces that traditional network security models were not designed to address.

My survey — currently under review at ACM Computing Surveys — systematically maps the landscape of side-channel attacks specific to SDN environments.


Three Attack Patterns That Stood Out

Control-plane timing leaks. Because SDN switches consult a centralized controller for flow-table misses, the round-trip time of a first-packet lookup leaks information about whether a flow rule exists. An attacker co-located on the data plane can fingerprint traffic policies or infer network topology without breaking any encryption.

Flow-table capacity exhaustion as a covert channel. TCAM-based flow tables have bounded capacity. A malicious tenant can flood the switch with synthetic flows, observe the degradation in legitimate traffic, and extract a timing signal correlated with other tenants’ behavior.

Controller load as an oracle. High-rate packet-in events saturate the controller’s processing queue. A patient attacker can modulate their own traffic to probe this queue and infer information about concurrent flows.


Attack Surface Summary

Attack Class Mechanism Information Leaked Mitigation Status
Control-plane timing First-packet RTT Flow-rule existence Partial
Flow-table exhaustion TCAM flooding Cross-tenant traffic Open problem
Controller load oracle Packet-in flooding Concurrent flow state Partial
Topology fingerprinting Probe packet timing Network topology Limited

Defense Gaps

The survey identifies three areas where no satisfactory mitigation exists yet:

Defense Approach Covers Gap
Randomized flow-table eviction Timing leaks Does not address capacity exhaustion
Rate limiting on packet-in Controller load oracle Can degrade legitimate traffic
Control-plane isolation Cross-tenant inference Incomplete in multi-tenant deployments

The survey also reviews proposed defenses and identifies gaps where no satisfactory mitigation exists yet. If you are working on SDN security or side-channel defenses and want to discuss the findings, feel free to reach out.




Enjoy Reading This Article?

Here are some more articles you might like to read next:

  • Building a Memory Leak Benchmark for Android with Soot
  • From IS Auditor to PhD Researcher: Why Industry Experience Changed How I Do Security Research