Theory Seminar

28 Feb
Wednesday, 02/28/2018 12:20pm to 1:20pm
Computer Science Building, Room 150
Theory Seminar
Speaker: Abhishek Agarwal

Distributed storage systems for large-scale applications traditionally use replication for data reliability. Error-correcting codes can be used in these systems to reduce the large storage overhead, while maintaining data reliability. A main limitation of traditional error-correcting codes is their high-repair cost during single node failure. In this talk, I will explore issues related to trade-off between repair-cost and storage overheard i) trade-offs for small alphabet constructions of codes for storage networks ii) trade-offs for construction of perfectly secure codes

In the second part of the talk, we will analyze schemes for a related problem of Index Coding : multi-casting data to a set of users each of which have side information about other users.