Full Program
Summary:
Nowadays heterogeneous systems architectures are becoming increasingly popular for running complex and heavy-load computational tasks. Many companies own vast data centers consisting of thousands of heterogeneous systems allowing developers to run their applications in GPUs FPGA or TPUs. However the rise in the use of such systems has led to a new vast attack vector with various possibilities to emerge. GPU vulnerabilities and their exploitations are realistic instances of such attacks. An instance of an attack on a GPU is to misuse a miner by continuously executing it and as a result overload the GPU and take it out for a long time period. This work presents a novel method of detecting attacks in heterogeneous systems by monitoring the PCIe traffic at a low level from the host to the corresponding PCIe endpoint.Author(s):
Iasonas Georgakas
Greece
Eva Papadogiannaki
TUC
Greece
Kostantinos Georgopoulos
TUC
Greece
Sotiris Ioannidis
TUC
Greece