Suboptimal feedback control of TCP flows in computer network using random early discard (RED) mechanism

We consider a dynamic model that simulates the interaction of TCPsources with active queue management system (AQM). We propose amodified version of an earlier dynamic model called RED. This isgoverned by a system of stochastic differential equations drivenby a doubly stochastic point process with intensity as thecontrol. The feedback control law proposed observes the router(queue) status and controls the intensity by sending congestionsignals (warnings) to the sources for adjustment of theirtransmission rates. The (feedback) control laws used are ofpolynomial type (including linear) with adjustable coefficients.They are optimized by use of genetic algorithm (GA) and randomrecursive search (RRS) technique. The numerical resultsdemonstrate that the proposed model and the method can improve thesystem performance significantly