Associate Professor Darshika G. Perera, Ph.D., and two of her graduate Ph.D. students, Mokhles Mohsin and Ahmed Alrasasi, and colleague S. Navid Shahrouzi, Ph.D., are innovating at the network’s edge, creating a framework for future research in next generation (next-gen), edge computing platforms.
The research, titled “Composing An Efficient and Adaptive Framework for Real-Time Processing on Next-Gen Edge-Computing Platforms: Models, Architectures, Methodologies, and Prototypes” is pushing boundaries in an area few people know about and understand.
Edge Computing
In the IoT era, cloud infrastructure alone is not enough to process and analyze the enormous amount of data being generated from various sensors and devices distributed throughout networks, such as smart grids.
Edge computing is emerging as a solution to address the issues associated with cloud computing. Edge computing is a distributed computing model where computing and data processing happen at the source in the network, like IoT devices, instead of in a traditional central cloud location. This model increases capabilities and efficiencies, allowing for immediate computations and data processing; however, edge computing is still in its initial stages.
Applications running at network edges are becoming more complex, thus requiring more processing power. Innovative solutions are needed to support the data-intensive applications on next-gen edge-computing platforms.
Perera’s main objective is to create an efficient and highly adaptive framework, including models, architectures, methodologies and prototypes, to support and accelerate real-time processing on next-gen edge-computing platforms. The National Science Foundation (NSF) funded this research.
Findings
Thus far, findings have been published in one journal paper in IEEE Access and an IEEE Canadian Review magazine article. Perera is authoring and submitting another IEEE Access paper as a second sequel.
The first paper introduced three different principal component analysis and support vector machine (PCA+SVM) models for real-time processing and analysis (for online training and inference) on edge-computing platforms.
Models 1 and 2 were created utilizing the same SVM algorithm, but with different design/functional flows. Model 3 was created with the same functional flow as Model 2 but utilized a modified SVM algorithm.
Experiments demonstrated that Model 3 was the most suitable computation model for real-time processing and analysis of edge computing platforms as compared to the other two models. Model 3 utilized a significantly lower number of iterations to produce results, while achieving acceptable performance results.
Continuing on with Model 3, in Perera’s second paper, a field-programmable gate array (FPGA)-based high-level-synthesis hardware architecture was introduced.
Impact
This unique and adaptive framework for real-time processing of computing and data-intensive applications, including data analytics and data mining, dramatically reduces the communication overhead and response delay of the networks and cloud infrastructure, while enhancing the performance and scalability of these systems.
The adaptive framework can be utilized both on the edge nodes and on edge devices and will prolong the useful life of systems while reducing the overall cost. Due to the reconfigurable nature of the designs, solutions can be configured for other tasks and will not be limited to one specific task. This will allow the framework to be utilized for a broad range of edge applications.
Inspiration
This research project is inspired by future trends and applications, such as smart grids, smart homes and autonomous vehicles, as well as space and remote uses. Many of these future applications require processing and analyzing the data closer to the source of the data, instead of sending data directly to centralized cloud data centers.
Cloud computing faces many challenges when transmitting, processing and analyzing big data, including insufficient bandwidth and communication delays, unsatisfactory real-time response, high power consumption and privacy protection issues.
Edge computing would address these issues by 1) pre-processing data at the edge, addressing insufficient bandwidth issues, 2) processing data in close proximity, enabling real-time processing and reducing delays, 3) by utilizing energy-constrained devices, reducing power consumption and 4) by not transmitting raw data, enhancing networks’ privacy/security.
Perera’s research is also inspired by her expertise in embedded/digital systems and reconfigurable computing, which will enable her to create an efficient and adaptive framework for resource-constrained edge devices and platforms.
About the UCCS College of Engineering and Applied Science
The College of Engineering and Applied Science enrolls more than 1,700 students and offers 24 engineering and computer science degrees, ranging from bachelor to doctoral. The college is a Department of Homeland Security / National Security Agency Center of Academic Excellence in Cyber Defense and works closely with the National Cybersecurity Center and with more than 250 aerospace and defense, information technology, cybersecurity and engineering organizations in the Pikes Peak region. Learn more about the College of Engineering and Applied Science at UCCS.