Towards Characterization and Optimization for Blockchain Systems

PhD Thesis Proposal Defence


Title: "Towards Characterization and Optimization for Blockchain Systems"

by

Miss Yuechen TAO


Abstract:

Blockchain technology has been widely used in a variety of distributed 
systems beyond cryptocurrencies, such as insurance, healthcare, and supply 
chains. However, it is well known that the performance in blockchain can 
suffer due to serialized transaction confirmations and potentially long 
consensus latency. In order to enable blockchain technology to be more 
accessible, this proposal studies the problems of performance 
characterization and optimization in several practical blockchain systems. 
Specifically, we first consider a sharded blockchain system designed for 
parallel transaction confirmations, in which existing solutions cannot 
achieve full parallelism as they incur a significant number of cross-shard 
validations and excessive communications. Second, we examine payment 
channel network (PCNs), in which transactions are confirmed instantly in 
an off-chain manner. We observe there is a fundamental tradeoff between 
the stringent privacy requirement and performance limitations.

Our first work focuses on improving the degree of parallelism in sharded 
blockchain systems by eliminating cross-shard validations with minimum 
cross-shard communications. This turns out to be challenging due to the 
transaction dependency and imbalanced sharding distribution in term of 
number of transactions in each shard. Specifically, transaction dependency 
affects cross-shard validations and imbalanced sharding distribution 
impairs the confirmation parallelization. We first study the inherent 
relationship between transactions and smart contract, and derive a 
sharding formulation based on smart contract. We then propose an 
inter-shard merging algorithm and an intra-shard transaction selection 
mechanism to minimize the imbalanced sharding distribution.

Our second work characterizes the fundamental performance limits in PCNs 
and examines the key factors involved. We develop a novel mathematical 
model capturing the PCN performance, and we study the impact from several 
key factors in particular the channel capacity and type of transactions. 
Through mathematical analysis, we derive the gap between the theoretically 
optimal performance and the performance achievable in practice, which 
helps to characterize the design space in PCNs for scheduling 
transactions.


Date:			Tuesday, 21 December 2021

Time:                  	3:00pm - 5:00pm

Venue: 			Room 1409
 			Lifts 25/26

Committee Members:	Prof. Bo Li (Supervisor)
  			Dr. Wei Wang (Chairperson)
 			Prof. Qiong Luo
 			Dr. Yangqiu Song


**** ALL are Welcome ****