Randomized Distributed Garbage Collection

With the explosive growth of the Internet and distributed computing, disributed garbage collection (DGC) has become a very important area of research. Single-processor garbage collection is by itself one of the challenging areas of research in computer science. In distributed environments, the problem of garbage collection becomes even harder and more challenging. This is mainly due to the additional overhead that is incurred by network communication and global sychronization. While most research efforts in DGC algorithms have focused on designing deterministic and correct algorithms, this work is a first attempt to introduce randomization techniques to DGC. We propose two approaches that could be applied in active objects systems and one approach that could be applied in actors system. Our mathematical analysis has shown that randomization acheives faster algorithms without highly impeding the accuracy and correctness of garbage collection.