RISE CoS A high performance Coproce Ssing Sensor

RISE & Co-S: A high performance Co-proce. Ssing Sensor architecture for offloading sensing and data processing Anirban Banerjee, Abhishek Mitra, Walid Najjar, Demetrios Zeinalipour-Yazti, Vana Kalogeraki, Dimitrios Gunopulos Department of Computer Science and Engineering, University of California, Riverside. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Roadmap n n n n n Problem Addressed Motivation Contribution Sense and Store RISE: Storage RISE: Processing Experimental Results Conclusions Future Work IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Problem Addressed n To design a sensor platform satisfying the following constraints: ¡ ¡ Small form factor Energy frugal operation High computational capability Gigabyte scale data storage IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Motivation n The need for a powerful sensor to match requirements for CCB, UCR. ¡ Ecological monitoring n ¡ Live species monitoring n n CO 2, Temperature, Humidity Bird call patterns using spectrograms There is a clear need for a sensor platform which can balance both energy frugalness and computational capability. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Contribution n n Developed the RISE (RIverside SEnsors) platform. Supplement the RISE with a Co. S (Coproce. Ssing architecture). Propose and implement “Sense and Store” paradigm. Developed Gigabyte-scale Storage Board with indexing capability. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

RISE Co-S Architecture IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Sense and Store n n n Pre-defined queries, e. g. max, min, top -k, avg…. Store queried values. Retrieve Queries from the index. Archive Data for offline processing. NAND Flash memory for bulk data NOR Flash memory for index IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Sense and Store (Motivation) Comparison of the Energy expended while transferring data using wireless and various storage options. Graph depicting the Energy expended for a particular Store Vs Send ratio. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Sense and Store (Motivation) 128 POINT FFT BENCHMARK RESULTS IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

RISE: Storage n Design / Features ¡ Upto 2 GB SD-Card / NAND n ¡ 1 MB on-board byte programmable NOR n n n Data archival, avoids expensive erasure operation throughout lifetime of node Index stored on NOR Flash Avoids need to preserve it onto the volatile SRAM Software Driver ¡ SPI Bus Interface n n ¡ ¡ SD-Card NOR Flash Archival of sensory data Fast retrieval / search using index Updating the index Updating Top-k values in the index IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Local Access Methods n n n Flash Card stores temporal data [ (ts, val 1, val 2, …. ), …, (ts, val 1, val 2, …. ) ] Example Query: “Find the timestamps on which val 1 (the temperature) was 95 F? ” Naïve Solution: Linear Scan over all pages on flash Card. Problem: Too expensive to search all pages. E. g. a 256 MB Flash Card features 500, 000 pages! Our Solution: Micro. Hash - which is an access method (index) deployed directly on the flash card to assist query evaluation. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Micro. Hash Overview n n An efficient access method that provides random and sorted access to records stored on the flash medium. Data and Index Pages are sequentially stored on Flash in a heap (sorted by timestamp). A buffer in SRAM / NOR keeps in memory the directory, index and data pages with the highest hit ratio. Micro. Hash structure provides a built-in wear leveling mechanism (evenly distributes page writes) and also minimizes number of expensive random page deletions. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Read Performance Improvements n n n Pages are read in 512 B blocks When pages are not fully occupied (i. e. index pages) then a lot of padding is transferred to SRAM. Since read can be performed at any granularity, we perform a Two-Phase Page Read (2 PPR). ¡ ¡ n 2 PPR - Phase 1: Read 8 byte Header and determine useful size of page. 2 PPR - Phase 2: Read in SRAM the useful bytes identified in Phase Note: To initiate a read we need a 9 -byte SD-Transaction. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

RISE: Processing n n n Reduce Phonetic renderings of bird songs into unique harmonics 128 point FFT on 16 -bit data sampled at 10 Ksps Calculate power spectral density Indexing the occurrence of harmonics USART interface from the M 16 C to the Chipcon CC 1010 host IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

FFT / Harmonics Extraction IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Energy Graph – FFT on Node Energy required to sample, Calculate FFT, Extract, Store and Transmit Harmonics, as compared to transmitting raw audio samples IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Conclusions n n n Developed RISE - Co. S: a powerful, energy frugal platform. Propose and Implement “Sense and Store” paradigm. Developed Giga-scale STB for in-situ data logging. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Future Work n n n 32 -bit, ARM-7 u. C onto the Co. S. (Preliminary results are encouraging) Integrate Microscopic, sub-surface camera equipment onto RISE. Make use of on-chip, security features of SD Card. IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra

Thank You Anirban Banerjee, Abhishek Mitra, Walid Najjar, Demetrios Zeinalipour. Yazti, Vana Kalogeraki, Dimitrios Gunopulos (anirban, amitra, najjar, csyiazti, vana, dg) @cs. ucr. edu IEEE SECON-2005, Tue, September 27, 2005 (C) Anirban Banerjee and Abhishek Mitra