1 Pico-IoT for a Smart NationProposers’ Day Pico-IoT for a Smart Nation Prof Dennis Polla Programme Director (Special Projects) scei.a-star.edu.sg 4 April 2017 Good morning and welcome to the A*STAR Proposers’ Day for Pico-IoT for a Smart Nation. Just a little over 1 year ago A*STAR SERC introduced a new concept called a Proposers’ Day to announce new larger-scale, longer-term thematic programs. The concept of a Proposers’ Day concept is used to ensure targeted outcomes in eventual research proposals to be submitted by the proposers community. The Proposers’ Day begins the process of soliciting new research ideas, in the form of a Call for Proposals, or LOIs, and attempts to ensure impact which for A*STAR which means eventual economic value capture for Singapore. I’m Dennis Polla and serve as the Director for Special Programmes at A*STAR. I am here to describe our interest in a possible Pico-IoT for a Smart Nation program. Call for LOIs (see slides #17-18) Announcement Date: 4 April 2017 Closing Date: 8 May 2017

2 Purpose and Agenda Purpose AgendaFamiliarize the research community with A*STAR’s interest in a thematic programme on “Pico-IoT for a Smart Nation” and define the scope of interests relevant to the programme to be started in late 2017. Outline the process and schedule for an A*STAR Scientific Workshop and and subsequent call for research proposals. Solicit feedback and questions. Agenda A*STAR’s interest in “Pico-IoT for a Smart Nation” Plan for a Scientific Workshop to be held on 18 May 2017 Programme Plan The purpose of the Proposers’ Day is to communicate A*STAR’s interests to potential research performers. We do not want to tell you what to propose – we want your own ideas We do have some areas of interest for our programme which might not be completely unique to A*STAR or to the Singapore research landscape. We are foremost concerned about carrying out research in Singapore that may have eventual economic impact. Of course, we want to do excellent science but eventual economic impact for Singapore is of primary importance.

3 Networked IoT for a Smart NationThe Internet of Things (IoT) is the internetworking of physical devices, vehicles, buildings and other items — embedded with electronics, software, sensors, actuators, and network connectivity that enable these objects to collect and exchange data. Singapore among other countries has recognized the importance of IoT and the need for cities (nations) to operate more efficiently. The need for Singapore to be a smart nation was announced 2 years ago and many challenges are now starting to take on even greater importance. For instance, millions of sensors are envisioned to be part of a smart city network, yet no one has really thought about the infrastructure cost of having millions of physical and chemical environmental monitors, personal monitors, and mobile platforms scattered throughout an urban environment, continuously collecting data and communicating information. We somewhat understand the advantages of batch-manufacturing of sensors using well established IC-type processes, but few have given thought to the problem of massive infrastructure management and maintenance of a large urban environment of people, traffic, noise, pollution, etc.

4 Proposers’ Day Purpose and AgendaThe Need Proposers’ Day Purpose and Agenda The concept of Unattended Ground Sensors (UGS) is not a new one. People have thought about it for almost 2 decades and have always talked about a mythical network of repeaters to transmit info through multiple nodes to a single processing site. In 2014, however DARPA re-examined why this concept doesn’t work – quite simply it is impractical to maintain the infrastructure needed for large-area info collection and transmission AND it does not provide rich information. But things have now changed … - We now have MEMS - We have low-power electronics - We now have very small antennas in every cell phone - We have much better batteries Maybe it is time to take another look at things. DARPA said what if we could make our defense and security sensors operate for the lifetime of a watch battery (and beyond)?

5 Sensors, Power and SignalThe Need Normally sleeping high power sensor Physical and Chemical Signatures Sensors, Power and Signal Processing Communication of Information Watt mWatt μWatt nWatt pWatt nm3 μm3 mm3 cm3 Conventional IoT Chip- IoT Pico- IoT Biological cells Organelle s Battery limited life Deeply-Scaled Systems Integration So DARPA launched the N-Zero Programme under the direction of Troy Olsson, but missed two important things: 1) The Internet-of-Things 2) Smaller is Better First, the IoT would be a potential winner if one could have sensors operating with < 1nW power (hence our choice of the name Pico-IoT) Second, DARPA-type defense and security needs don’t always need to be small scale solutions, and hence Near-Zero Power suitcases deployed as collection sensors in environment are allowable. But in really studying the problem, it becomes very hard to simul-taneously scales both power and size to sub-mm and sub-nW levels. So, we can learn a lot from DARPA N-Zero, our program has commercial IoT in mind at a reasonable cost enabled by batch-manufacturing. And we follow Feynman: “There’s plenty of room at the bottom.” This summarizes the need: We want “Pico-IoT” which are sensing, computing, and communication devices and systems that consume very low power (<1nW) and occupy minimal volume (<1mm3). Size [Diagram was prepared for A*STAR by Prof Amit Lal, Cornell University, March 2017 ] Power

6 Hard Challenges 3) Low-Power Signal 4) Low-Power Communication[Physical and chemical sensors that hibernate with near zero power (<1 nW) consumption until woken-up] [There are many energy scavenging principles, but practical energy harvesting from ambient sources has not been demonstrated.] 1) Ultra-Lower Power Event-Driven Sensors 2) Energy Extraction from Ambient Sources [Low-power digital circuits have been developed to crunch numbers in mobile applications, but the equivalent node analog circuits (sub-threshold CMOS) are far away.] [Integrated communication linkages for MEMS is an open research field. Size, duty cycle, comms distance, frequency, and operation may be different than that used in commercial handphones.] Here are the hard challenges … Each of these challenges are briefly summarized in the following charts. 3) Low-Power Signal Analog and Digital Processing 4) Low-Power Communication Linkages

7 HARD Challenges 3) Low-Power Signal 4) Low-Power Communication[Physical and chemical sensors that hibernate with near zero power (<1 nW) consumption until woken-up] [There are many energy scavenging principles, but practical energy harvesting from ambient sources has not been demonstrated.] 1) Ultra-Lower Power Event-Driven Sensors 2) Energy Extraction from Ambient Sources [Low-power digital circuits have been developed to crunch numbers in mobile applications, but the equivalent node analog circuits (sub-threshold CMOS) are far away.] [Integrated communication linkages for MEMS is an open research field. Size, duty cycle, comms distance, frequency, and operation may be different than that used in commercial handphones.] Emphasis Added: I do not want to underestimate the difficulty and the fact that this is high-risk. 3) Low-Power Signal Analog and Digital Processing 4) Low-Power Communication Linkages

8 HARD Challenges 5) Systems Integration, Extreme[Truly small systems with sub-mm features consisting of sensors, electronics, communication links, and comms have never been realized in a small volume. 5) Systems Integration, Extreme Miniaturization, Heterogeneous Integration

9 Hard Problems 1) Ultra-Lower Power Event-Driven SensorsPico-IoT for a Smart Nation is particularly interested in persistent, event-driven physical and chemical sensing capabilities where IoT systems can remain dormant (consuming very little or no power) until woken by an external stimulus or triggering event. Event-Driven sensors must be able to immediately respond to the stimulus and perform an intended sensing operation(s). After reporting their result, event-driven sensors are subsequently required to return to a dormant state until the next triggering event. Such continuous and passive operation precludes systems designs that are one time single event-driven2. New sensors that can demonstrate very low power consumption, or sensors that generate energy directly from the ambient source signal, and still generate high signal-to-noise ratios are of high interest. 2) Energy Extraction from Ambient Sources Pico-IoT must be able to operate for their intended design lifetime without the need for battery replacement. For sensing applications in remotely accessible locations or permanently embedded in structures or buildings, operating lifetimes may be over 10 years and require an ability to extract energy (electromagnetic, acoustic, thermal, chemical, etc.) from the ambient surroundings, store this energy, and deliver power when required. New sensor approaches that harness signals as a source of both energy and information with pW to nW power levels are of interest.

10 Hard Problems 3) Low-Power Signal ProcessingMeaningful information must be acquired and processed electronically. This requires electronic circuitry which also may require very little or no power when dormant and operate within a minimal power budget set by the overall systems constraints. Low-voltage, sub-threshold circuits for analog signal processing are of interest. Compatible interfacing with sensors and communication linkages is required. Subthreshold circuits with built-in semiconductor and nanomechanical switches that also consume pW to nW of power are encouraged. 4) Low-Power Communication linkages RF and other modes of communications such as optical and ultrasonic communication linkages are required for reporting events. Such communication linkages and data transfer must be designed for both periodic and infrequent communication with receivers commonly emplaced in a dense urban environment. Approaches that demonstrate very low power communication channels between IoT systems are encouraged. 5) Systems Integration, Extreme Miniaturization, Advanced Heterogeneous Integration The continuous and autonomous sensing operations based on millions of sensors are needed for future dense urban environments. Issues of size, maintenance over a design lifetime, and systems compatibility are important for each type of IoT sensor. Ideally, full systems cost and associated cost of ownership over the design lifetime are important requirements. Overall architectures and anticipated protocols for operation must be considered.

11 Relevant Research Troy Olsson, DARPA Event-driven security needs mayWe clearly want to acknowledge the DARPA N-Zero Program. Event-driven security needs may operate on relevant principles needed for massively networked IoT applications. Academic research performers in “N-ZERO” are unknown but may soon have relevant publications in the open literature.

12 Frank Gerlach, Technical University of DresdenRelevant Research Frank Gerlach, Technical University of Dresden Binary-Zero-Power Sensor for Autonomous Networks The paper by Gerlach, et al., from the Technical University of Dresden provides an early research exploration of the Zero-Power Sensor concept. AND I am sure your will find a wealth of information on the state-of-the art for the 5 problem areas presented. Few however have thought about putting all 5 together which is the objective of the Pico-IoT Programme. Principle of Binary-Zero-Power Sensors

13 What Problem Are You Trying to Solve?Current IoT sensors are limited by both their size and power. Because millions of IoT sensors are needed for improving efficiencies and providing increased safety in urban areas and structures, simultaneous attention to size, power, and cost of ownership will be important considerations for ubiquitous sensing and monitoring multiple variables at multiple collection pointys. In particular, the cost to access and change batteries has become a key challenge in both urban and remote environments.

14 Scope of A*STAR’s InterestSolve the hard problems holding back Pico-IoT for a Smart Nation: - Ultra-Lower Power Event-Driven Sensors - Energy Extraction from Ambient Sources - Low-Power Signal Processing - Low-Power Communication linkages - Systems Integration, Extreme Miniaturization, Heterogeneous Integration Establish scientific and technology capabilities in Singapore that attract other researchers from around the world Advance the Singapore Smart Nation Initiative But to get there, we don’t want to just build capability only. We want to solve hard challenges. Your response does not need to solve all 5 problems. A nice research proposal addressing just one or more of them is what we are now requesting. We will likely look for synergies and could develop one holistic program. This will be decided based on the responses received and scientific workshop.

15 Call for Research IdeasWhat problem are you trying to solve? How do you intend to solve the problem? What milestones do you plan to use to ensure progress? 4. How do you intend to quantitatively measure your research success? Examples of Interest New standby sensing approaches New energy harvesting methods (e.g. energy harvesting of ambient rf energy) Sub-threshold circuits with near-zero standby power Low power rf links designed for sensors and infrequent event-driven reporting Extremely miniaturized integrated systems and packaging We want to develop great proposals with you and we want to begin a dialogue with you starting with your idea communicated to us in the form of a pre-proposal (LOI).

16 Not of Interest Research that only builds capability and infrastructure Research ideas that are not directed toward solving a hard problem Research goals which are not quantitatively specified and measured Incremental research approaches that merely build upon solutions that are already known and demonstrated, e.g. 20% improvement in metric x (e.g., solar cell approaches). Engineering-only solutions with no high risk research unknowns Ideas specifically directed toward applications and short-term commercialization Supplements to active research already underway Mention of traditional KPIs including number of students trained, number of publications, number of research citations, etc.

17 Next Steps Identify the hard problem you are trying to solveCall for LOIs Announcement Date: 4 April 2017 Closing Date: 8 May 2017 Identify the hard problem you are trying to solve Submit a 9-11 page ppt presentation in ppt format by 3:00 pm 8 May 2017 (Only submit a pdf version if your ppt file cannot be compressed below 25 MB. You will be contacted later for submitting the ppt version) Required Response to be sent to 1. Title slide with name(s), affiliation, and contact info 2. Background description of the state-of-the-art 3. Statement and description of the problem to be solved 4. Key technical approach to be undertaken (1-2 slides); be sure to describe your new insight and thinking that others may have missed 5. Major challenges your technical approach will address; identify why this is a challenging research project with unknowns and why it is hard 6. Organization’s capabilities and PI qualifications; a concise summary of relevant past research accomplishment (optional) 7. Summary with technical endpoint or final quantitatively measured goal identified including your proof-of-concept, platform demonstration, or prototype deliverable. 8. How will you quantitatively measure success of the research to be carried out. 9. Anything else you want to add; do not specify cost or usual KPIs 10. Institutional acknowledgement letter #8 Prepare your own slide to tell us how you want to be quantitatively measured if you are on-track and have been successful in carrying out your research activity.

18 Evaluation Criteria a) Relevance to the Programme Goals (15%)Addresses one or more stated challenges in the research programme. (5 points) Proposed research is deeply relevant to the research programme. (10 points) b) Technical Approach (45%) Describes the state-of-the-art in the field. (10 points) Approach is innovative and different from what others in the scientific community are already pursuing. (25 points) Approach does NOT represent incremental improvement over other known solutions. (10 points) c) Potential Research Impact (30%) Proposed research, if successful, would draw significant interest from the scientific community. (10 points) Proposed research, if successful, would have eventual relevance to industry. (10 points) Proposed research includes plans for a well-described deliverable or prototype of some sort. (10 points) d) Proposer Capabilities (10%) Past high-impact work in this area (journal publications, patents, prototypes). (5 points) Assembled a multi-disciplinary team for the research. (5 points) Total: nts Some Pre-Proposals will be selected for presentation at an A*STAR Workshop to be held on 18 May 2017. The workshop will be open to Singapore researchers and a few invited international researchers who will have signed NDAs. Technical presentations at this workshop should not disclose truly proprietary information such as specific recipes, etc. If you believe a presentation in such an open workshop environment is not possible you may choose to submit your idea through any of the other open-call channels which A*STAR has already announced.

19 Programme Management Launch of a portfolio of research approaches which takes 3-5 years to complete. Base period funding for 12 months. Options are exercised based on satisfactory performance and relevance to the current scientific state-of-the-art. Possible discontinuation and consolidation of approaches; teaming Contract modification when necessary Periodic reporting of research progress Quarterly review of research performance. Researchers share their results with one another periodically in a workshop venue These are details that will be worked out later.

20 “Pico-IoT for a Smart Nation” PlanTentative Timeline (subject to change) Announcement Date: 4 April 2017 Closing Date: 8 May 2017 Pre- Proposal Eval (12 May) Advisory Panel Mtg. (19 May) Evaluation/ Endorsement Panels (as required) Programme Planning Workshop Planning Programme Planning Approvals Proposers’ Day Announcement (23 Mar) Proposers’ Day (4 Apr) Pre- Proposals (LOIs) Due (8 May) Scientific Workshop (18 May) Proposal First Draft (15 Aug)

21 Thank you