Supporting Resiliency and Timeliness in Edge Applications with Dispersed Computing
ABSTRACT
In mission-critical applications that rely upon edge components, any dependence on centralized data centers and shared networks can influence the overall efficiency and efficacy of the application. Losses of links, congestion, or jitter during data delivery, and over-extended CPU and memory are examples of faults and stresses that may cumulatively degrade performance. In this talk, we describe a current research effort that is investigating and prototyping transparent middleware-based approaches for utilizing emerging in-network compute resources to migrate and host computation in optimal locations with respect to demand and resource availability. We then suggest how such capabilities can improve the performance and resiliency of IoT applications.
BIO
Mr. Aaron Paulos is a principal investigator and computer scientist at BBN Technologies. At BBN, he investigates transparent adaptive computing paradigms as a method to improve the security, resiliency, and trust properties of distributed applications. Over the last decade, Aaron has supported a number of AFRL, DARPA, and IARPA research efforts. Prior to joining BBN, he researched fault tolerance and assistive technologies for the blind at the ECE department of Carnegie Mellon University. Aaron has an M.S. from the Information Network Institute at CMU and a B.S. in Computer Science from the University of Pittsburgh. He is also a member of the ACM and IEEE.