Trustworthy Online Social Media
The prosperity of online social networks makes it much easier and more convenient for individual users to proactively generate, share and exchange diverse digital content online. With more people getting used to make their economic, political, and even daily life decisions by referring to information from online social media, most of them are not aware how their decisions are actually influenced by malicious attackers, companies or politicians through fabricating and propagating false information on OSNs.
My research in this direction aims to protect a healthy and trustworthy online social environment where trustworthy information can be boosted while dishonest and hostile information is suppressed. Specifically, to address the information manipulations and ensure the security, privacy and trust in OSNs, in this line of research, we mainly utilized machine learning algorithms and probability-based signal processing techniques to perform false information detection. Meanwhile, we also studied information manipulations and privacy attacks from the attacker’s point of view, with the ultimate goal as to help resolve the vulnerabilities of current defense schemes. This research stream is significant because security, privacy, and trust issues on online social media are closely related with everyone’s daily lives and attracts increasing attention from both academia and industry.
Ensure trustworthy computing and communications on resource-constrained IoT edge devices
With the growing prevalence of the Internet of Things (IoT), securing the sheer abundance of devices is critical. More importantly, many of these IoT devices are battery-based devices with very limited computing resources, such as CPU, memory, and storage, making it challenging to afford the high computational requirements from conventional security solutions.
In this line of research, we have successfully established a WiFi based comprehensive IoT testbed with real devices. Based on this testbed, we have (1) quantitatively evaluated the energy consumption of edge devices when they execute the state-of-the-art cryptographic algorithms and security protocols; (2) analyzed the impact of DDoS and energy based DDoS attacks on the edge devices; (3) proposed lightweight features for machine learning algorithms to perform malicious traffic detection; and (4) explored effective approaches to offload computation/communication overhead from edge devices to trustworthy fog nodes. The adoption of real testbed differentiates our work from most existing studies, making our research outcomes particularly practical and transformative.
Trustworthy occupancy detection model from economic sensors for building energy saving
Buildings are responsible for 70% of electricity consumption, and 48% of the total energy consumption in the US. Recent advancements in Cyber-Physical Systems (CPS) expanded opportunities to engage more people into energy-saving practices. However, most of the current functions of available smart home products are limited to remote controllability of an energy device and lack of data-based automatic learning/controlling. In addition, the sensors involved in most of these products (e.g. camera or audio sensors)are expensive and intrusive, raising great consumers’ concerns oncosts and privacy. Therefore, the goal of this project is to achieve automatic energy saving by developing a trustworthy learning model to determine occupancy information in a residential building from multiple economic and non-intrusive sensors.
Archived Projects
Towards Trustworthy Neural-controlled Artificial Legs using High-Performance Embedded Computers
There are over 32 million amputees worldwide whose life are severely impacted. A continued need exists to provide this large and growing population of amputees with the best care and return of function possible. The quality of life of leg amputees can be improved dramatically by using a cyber physical system (CPS) that controls artificial legs based on neural signals representing amputees’ intended movements. The key to the CPS system is the neural-machine interface (NMI) that senses electromyographic (EMG) signals to make control decisions.
In this design, the security issue is very important, since any wrong decisions may lead to falls of the amputees, which is very dangerous. We apply trust model in this application to evaluate the reliability of the EMG sensors in real time, help the NMI system to make sensor removal and entering decisions and ensure the security of the entire NMI system.
Some experiment videos are available at