Hidehisa Takamizawa

Customizing Virtual Machine with Fault Injector by Integrating with SpecC Device Model for a Software Testing Environment D-Cloud

D-Cloud is a software testing environment for dependable parallel and distributed systems using cloud computing technology. We use Eucalyptus as cloud management software to manage virtual machines designed based on QEMU, called FaultVM, which have a fault injection mechanism. D-Cloud enables the test procedures to be automated using a large amount of computing resources in the cloud by interpreting the system configuration and the test scenario written in XML in D-Cloud front end and enables tests including hardware faults by emulating hardware faults by FaultVM flexibly. In the present paper, we describe the customization facility of FaultVM used to add new device models. We use SpecC, which is a system description language, to describe the behavior of devices, and a simulator generated from the description by SpecC is linked and integrated into FaultVM. This also makes the definition and injection of faults flexible without the modification of the original QEMU source codes. This facility allows D-Cloud to be used to test distributed systems with customized devices.

A replica management protocol in a binary balanced tree structure-based P2P network

The purpose of our work is to realize a load balancing of nodes in a P2P network. A replica management protocol, which exploits replicas for balancing loads of each node managing popular data, by adapting replica partition trees to a balanced tree overlay network called BATON, for BAlanced Tree Overlay Network, is proposed for this. BATON has a load balancing mechanism where each node adjusts the number of data managed by it. However, if there are some popular data that are accessed by a large number of nodes in a network, the mechanism of BATON could fail. We propose a replica management protocol for balancing loads of both data transmission and replica management of each node. Some results of simulation in which our proposal was compared with a method without replica and another method of simple replica management are showed and the effective and weak points of our proposal are discussed.

Applying Privacy Preserving Count Aggregate Queries to k-Classification

It is important to process data effectively while preserving privacy of personal information. In this paper, we propose a technique to reconstruct results of count aggregate queries from a perturbed table for building a decision tree whose target attribute contains more than two classes. Using the conventional technique, we must reconstruct the results of target values from those of each value calculated independently in such the case. In this paper, we borrow and extend the conventional technique to reconstruct the results of target values at once. We also report some experimental results showing that our proposal can reduce reconstruction errors compared to the conventional technique in cases where perturbation ratio is high.