New and emerging reality technologies, such as augmented and virtual reality, require fast data transmission and minimal delay in order to support interactive and immersive user experiences. However, achieving both high speed and low latency often leads to a high bit error rate, which reduces the dependability and quality essential for applications like video conferencing and 3D telemedicine.
With a new $125,000 grant from the National Science Foundation, School of Computing Assistant Professor Hongzhi Guo proposes a new error detection and correction coding solution that could potentially transform how communication systems process and overcome errors.
Bit errors that occur within communications systems possess varying levels of significance. Errors with low significance can usually be tolerated by the user, but those with high significance have a greater effect on performance and must be corrected. Guo compares bit errors to typos in text messages. While the recipient will likely still be able to understand a message missing only one letter or word, a typo that changes a word like “computing” to “commuting” could cause more confusion.
“Traditional bit-level error correction treats all bits as equally important, but not all bit errors impact the perceptual quality or structural integrity of the message,” Guo said. “Topological coding prioritizes preserving essential features that are critical for human understanding or machine interpretation.”
Guo’s project will explore extracting and encoding topological information from source data through Topological Data Analysis (TDA). This technique would ensure data fidelity at the topology level rather than the bit level, allowing for error detection and correction through recovery of the original topological information.
“Because it captures high-order structural information, this representation is more compact and semantically meaningful than traditional bit-level data,” Guo said. “Upon receiving the transmitted data, we compare its topological structure with the original. If discrepancies exist, we can detect the presence of errors and assess their significance based on the extent of structural deviation.”
Not only would Guo’s project offer new insights into next-generation communication system errors, but it would also enable the development of novel algorithms to streamline the error detection and correction process. Algorithms with the ability to evaluate the significance of errors could ultimately allow them to be corrected in real-time alongside other system optimizations.
While other coding solutions capable of distinguishing data meaning and error significance do already exist, most rely on deep learning models, which are typically application- and modality-specific, limiting their broad usage across various types of data and systems. Guo said he has been seeking a semantic coding approach that could ensure data fidelity without depending on training-based deep learning models, and his computational topology approach offers such a framework. He looks forward to future developments in this converging area of computing and communications.
“Today, with advancements in topological data analysis and persistent homology, we have many available mathematical tools to pursue this direction,” Guo said. “I’m excited to explore and contribute further to this promising intersection of computational topology and data communication.”