Abstract
Due to the success of C++ and Java，Object-oriented applications dominate the modern programming marketplace。As a result，the use of dynamic storage allocation in application programs increases dramatically。On such applications，modular programming and heap objects are extensively used。Garbage Collection is then introduced to support fully modular programming。Among traditional and recent Garbage Collection strategies，we are interested in the Generational Garbage Collector and the Lifetime predictor。After studying and learning related works，we describe and simulate a Profile-Based、Lifetime-predicted Garbage Collection with dynamic generation shifting。The proposed Garbage Collection strategy amends the drawbacks of traditional lifetime predicator and offers a scheme for solving write barrier problem in incremental tracing as well。