Intelligent SDN: Adaptive Traffic Engineering & Load Balancing

01 // SYSTEM OVERVIEW

An autonomous Software-Defined Networking (SDN) controller engineered in Python on top of the POX framework and Mininet emulation environment. Proactively monitors link utilization across complex network topologies, isolates bandwidth-heavy elephant flows, and dynamically installs OpenFlow rules to balance traffic across redundant paths.

02 // PROBLEM SPECIFICATION

Traditional static routing protocols (like OSPF or ECMP) direct traffic along rigid shortest paths, resulting in severe packet loss and link bottlenecks while alternative parallel paths sit idle.

03 // ENGINEERING SOLUTION

Designed an adaptive traffic engineering controller that periodically interrogates switch port statistics via OpenFlow. When port utilization crosses dynamic thresholds, the controller isolates elephant flows and calculates lowest-cost disjoint paths to re-route heavy flows without disrupting existing flows.

04 // SYSTEM ARCHITECTURE & DATA FLOW

Mininet Emulated Topology (Open vSwitch Instances) <-> OpenFlow 1.0 Protocol <-> POX Controller Core (Port Statistics Collector -> Flow Classifier -> Dijkstra Disjoint Path Calculator -> FlowMod Rule Injector).

05 // VERIFIED OUTCOMES

• [1]Eliminated single-link congestion under saturated test transfers using dynamic rerouting
• [2]Maximized aggregate network throughput across multipath topologies during concurrent Iperf benchmarks

06 // TECHNICAL CONSTRAINTS & CHALLENGES

• —Preventing route oscillation (flapping) when multiple heavy flows simultaneously compete for alternative paths
• —Minimizing OpenFlow controller inquiry overhead while maintaining real-time telemetry responsiveness

07 // ARCHITECTURAL TAKEAWAYS

• →Hysteresis dampening and adaptive utilization thresholds are essential to prevent route flapping in SDN controllers
• →Fine-grained packet counters in OpenFlow allow precise discrimination between elephant and mice flows