Network Footprint and Optimization for Secure Computation

PhD Qualifying Examination


Title: "Network Footprint and Optimization for Secure Computation"

by

Mr. Zhenghang REN


Abstract:

Datacenter is one of the most important facilities in the era of big data 
and artificial intelligence as it stores large datasets and provides 
high-performance computation clusters. However, storing large datasets in 
the datacenter risks privacy because the datasets may contain sensitive 
information of individuals. Privacy-preserving computation techniques such 
as Secure Multi-party Computation (MPC) and Hardware Enclave have been 
explored in machine learning and big data analysis to protect privacy. 
However, privacy- preserving techniques usually bring extra cost in 
communication, making them impractical, especially for complex workloads.

Current optimizations on secure computation usually focus on improving the 
efficiency of protocols or finding the optimal combination of protocols. 
However, the network facilities such as Remote Direct Memory Access (RDMA) 
and the upper-level software stack such as Remote Procedure Call (RPC) for 
communication are also critical for performance, especially for 
communication-intensive protocols such as secret sharing. Thus, exploring 
opportunities to accelerate communication in secure computation with the 
high-performance network is necessary.

In this survey, we explore the network characteristics and chances for 
optimization when performing secure computation protocol in datacenter 
network. We explain the gap between the secure computation 
applications/protocols and high-performance Network Interface Controller 
(NIC) and explore chances to optimize the performance. Moreover, we also 
listed and analyzed some challenges faced with characteristics of secure 
computation that need to be addressed.


Date:  			Tuesday, 30 August 2022

Time:                  	2:00pm - 4:00pm

Zoom Meeting: 
https://hkust.zoom.us/j/99681406149?pwd=cFBSREVPcnQzcXYrT2NhYXYwQkNWZz09

Committee Members:	Prof. Kai Chen (Supervisor)
 			Dr. Dimitris Papadopoulos (Chairperson)
 			Prof. Gary Chan
 			Prof. Ke Yi


**** ALL are Welcome ****