Ldu Parametrized Discrete-time Multivariable Mrac and Application to a Web Cache System

Aimed at developing a new resource allocation technique using adaptive feedback control for web cache systems which are inherently dynamic and have multiple inputs, multiple outputs and parameter uncertainties as well, the rst goal achieved in this paper is to develop a multivariable discrete-time model reference adaptive control (MRAC) scheme with relaxed design conditions based on an LDU decomposition of the system high frequency gain matrix. The developed MRAC scheme ensures global signal boundedness and asymptotic output tracking, as well as completely singularity-free control signal generation. The second achieved goal is the application of this new adaptive control scheme to a web cache system to provide proportional diierentiation on average hit rate of diierent content classes of client requests. A key task fulllled is system modeling. With the measured relative cache hit rates as outputs and the content class storage space ratios as inputs, a multivariable dynamic web cache model is derived and experimentally veriied via recursive system identiication using real system data for a 2-input 2-output case. Based on this model, the adaptive control scheme is implemented for feedback control of the chosen web cache. Experimental results are presented, which, as desired, show an evident promise of an eeective new approach of using adaptive control in web cache systems for performance guarantees in the presence of system uncertainties.