No abstract

Abstract. This paper introduces a novel technique for extracting the unique timing signatures of the FPGA configurable logic blocks in a digital form over the space of possible challenges. A new class of physical unclonable functions that enables inputs challenges such as timing, digital, and placement challenges can be built upon the delay signatures. We introduce a suite of new authentication protocols that take into account non-triviality of bitstream reverse-engineering in addition to the FPGA's unprecedented speed in responding to challenges. Our technique is secure against various attacks and robust to fluctuations in operational conditions. Proof of concept implementation of the signature extraction and evaluations of the proposed methods are demonstrated on Xilinx Virtex 5 FPGAs. Experimental results demonstrate practicality of the proposed techniques.

The growing popularity of multi-tenant, cloud-based computing platforms is increasing interest in resource allocation models that permit flexible sharing of the underlying infrastructure. This thesis introduces a novel IO resource allocation model that better captures the requirements of paying tenants sharing a physical infrastructure. The model addresses a major concern regarding application performance stability when clients migrate from a dedicated to a shared platform. Specifically, while clients would like their applications to behave similarly in both situations, traditional models of fairness, like proportional share allocation, do not exhibit this behavior in the context of modern multi-tiered storage architectures. We also present a scheduling algorithm, the Reward Scheduler, that implements the new allocation policy, by rewarding clients with better runtime characteristics, resulting in benefits to both the clients and the service provider. Moreover, the Reward scheduler also supports weight-based capacity allocation subject to a minimum reservation and maximum limitation on the IO allocation for each task. Experimental results indicate that the proposed algorithm proportionally allocates the system capacity in proportion to their entitlements.

The growing popularity of multi-tenant, cloud-based computing platforms is driving research into new QoS models that permit flexible sharing of the underlying infrastructure. In this paper, we reexamine the use of the popular proportional-share model for resource allocation, in the context of modern heterogeneous, multitiered storage systems. We highlight the limitations of the conventional proportional sharing approach to resource allocation, and describe a new allocation model that provides strong isolation between clients. This improves the performance characteristics from the viewpoints of both the clients and the service provider.