Communication channel analysis and real time compressed sensing for high density neural recording devices

摘要

Next generation neural recording and Brain-udMachine Interface (BMI) devices call for high density or distributedudsystems with more than 1000 recording sites. As theudrecording site density grows, the device generates data on theudscale of several hundred megabits per second (Mbps). Transmittingudsuch large amounts of data induces significant powerudconsumption and heat dissipation for the implanted electronics.udFacing these constraints, efficient on-chip compression techniquesudbecome essential to the reduction of implanted systems powerudconsumption. This paper analyzes the communication channeludconstraints for high density neural recording devices. This paperudthen quantifies the improvement on communication channeludusing efficient on-chip compression methods. Finally, This paperuddescribes a Compressed Sensing (CS) based system that canudreduce the data rate by 10x times while using power onudthe order of a few hundred nW per recording channel.