Database Compression for Security:● Self-design and build the entire system of a data compression approach for causality analysis, including the data connecting from Postgre Database, compression, and decompression, query on the compressed data, causality analysis upon the compressed database. ● Based on information-theoretic observations on system event data, the approach achieves high lossless compression and supports near real-time retrieval of historic events. ● Self-implemented experiments on two real-world data sets (DARPA and NEC lab data) show that our approach offers 2.63x and 12.94x data size reduction.LDPC code for DNA storage.● Self-implemented the whole LDPC code system, including encoding, decoding, parity-matrix design,density evolution design, etc. ● A designed asymmetric error model for both discrete and continuous sequencing output. The continuous output is based on the analog current signal from nanopore sequencing. ● Designed the binary LDPC code with a turbo-like decoder to correct the asymmetric substitution error during the DNA storage. ● Designed the density evolution algorithm to theoretically prove the coding error rate can reach the Shannon capacity. ● Simulation results show the improved LDPC code is 4X faster than the traditional quaternary LDPC code.Constrained codes & Viterbi decoding for DNA storage: ● Self-implemented the whole constrained code system, including encoding, decoding and state graph design, etc. ● Designed a substitution and deletion channel model based on k-mer nanopore sequencing model. ● Improved traditional constrained code for the quaternary code and present methods to reduce the time andspace complexity. ● develop a decoder based on the Viterbi algorithm that allows correction of additive Gaussian noise and deletions.

ENGR 7A and ENGR 7B Introduction to Engineering Design - QuadcoptersEECS 40 OBJ ORIENT SYS&PROG. Java based object oriented programming.

Clock synchronization in distributed systems.The study is on a low latency (around 100ns) distributed system with multi nodes.● Help team design support Vector Machine for the synchronization between one pair of nodes and reducethe latency from 100ns to around 10ns.● For the distributed system with multiple nodes, implemented the network effect algorithm that reduces the error variance.● Self-implemented the simulation program for the company to analyze the potential optimal networkconstruction for the model. The simulation considered the error variance and effect from nodes failure.