Protocols to Reduce CPS Sensor Traffic using Smart Indexing and Edge Computing Support

TY - GEN

T1 - Protocols to Reduce CPS Sensor Traffic using Smart Indexing and Edge Computing Support

AU - Yazdani, Niloofar

AU - Lucani Rötter, Daniel Enrique

PY - 2019/12

Y1 - 2019/12

N2 - We propose a new approach for lossless data compression to reduce the amount of data transmitted by Cyber-Physical Systems (CPS) by several-fold. Our approach uses an indexing technique inspired in the concept of generalized deduplication to compress data by a) finding matches of similar (not necessarily equal) data chunks, and b) exploiting data chunks previously stored in the Edge (or Cloud) by the same or even other CPS devices. We propose a mathematical model to predict the gains based on the number of previously received chunks and validate it using simulations. We show that compression factors of 23-fold are possible even with a limited number of previously stored chunks in the Edge. Gains of 2.8-fold over DEFLATE compression on differential samples are possible even for small data chunks (16 B) for synthetic data. Using real-world CPS data sets, we show that our technique can provide gains of up to 2.9. We also show our solution's processing speed in a Raspberry Pi 3 can be as high as 163 MB/s.

AB - We propose a new approach for lossless data compression to reduce the amount of data transmitted by Cyber-Physical Systems (CPS) by several-fold. Our approach uses an indexing technique inspired in the concept of generalized deduplication to compress data by a) finding matches of similar (not necessarily equal) data chunks, and b) exploiting data chunks previously stored in the Edge (or Cloud) by the same or even other CPS devices. We propose a mathematical model to predict the gains based on the number of previously received chunks and validate it using simulations. We show that compression factors of 23-fold are possible even with a limited number of previously stored chunks in the Edge. Gains of 2.8-fold over DEFLATE compression on differential samples are possible even for small data chunks (16 B) for synthetic data. Using real-world CPS data sets, we show that our technique can provide gains of up to 2.9. We also show our solution's processing speed in a Raspberry Pi 3 can be as high as 163 MB/s.

KW - Cyber-Physical Systems

KW - Data deduplication

KW - Data transmission reduction

KW - Wireless sensors

UR - http://www.scopus.com/inward/record.url?scp=85082296620&partnerID=8YFLogxK

U2 - 10.1109/GCWkshps45667.2019.9024686

DO - 10.1109/GCWkshps45667.2019.9024686

M3 - Article in proceedings

BT - 2019 IEEE Globecom Workshops, GC Wkshps 2019 - Proceedings

PB - IEEE

T2 - 2019 IEEE Globecom Workshops

Y2 - 9 December 2019 through 13 December 2019

ER -