Environmental Design for Pervasive Computing Systems Ravi Jain

Environmental Design for Pervasive Computing Systems Ravi Jain John R. Wullert II Autonomous Comm. Lab Do. Co. Mo USA Labs Applied Research Telcordia Technologies jain@docomolabs-usa. com jwullert@telcordia. com Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved.

Outline • Do. Co. Mo USA Labs • • • Motivation and background Using less Using it longer Smart disposal Conclusions Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 2

Summary • Do. Co. Mo USA Labs is doing exciting research on 4 G wireless services, architectures, and networks • Pervasive computing offers tremendous opportunities but also possible negative environmental impacts • We need to treat environmental impact as a first-class design constraint • Increasingly software is a key to reducing environmental impact • We need research in many areas of system design for pervasive computing and communications Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 3

NTT Do. Co. Mo • Parent: NTT Corp (inc. 1952) • History – 1979: NTT offered a car-phone service • 21 units in Tokyo metro, $3 K monthly charge, $10 K security deposit – 1992: Do. Co. Mo spun out • 0. 5 M cellphone subscribers, 8 M pager subscribers – 1999: Do. Co. Mo launched i-mode data service – 2001: Do. Co. Mo launched W-CDMA (FOMA), world’s first 3 G service Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 4

NTT Do. Co. Mo Market Trends Source: NTT Do. Co. Mo website: http: //www. nttdocomo. com, Feb. 2003 Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 5

3 G (FOMA) Application Plan New servi ce d ep l o ymen t International roaming M-stage visual on FOMA M-stage music on FOMA Image-clipping with i-mode ＴＶ Phone Mobile EC Music Delivery i-motion ｉ-mode Video Delivery Location information Visual mail Possible to use FOMA and 2 G phones with a single phone number Send images instantly as mail Dual network service Han d set en h an cemen t PDA type Upon service launch 2 G/3 G dual phone Smaller, lighter handsets with longer battery time Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 6

Do. Co. Mo’s R&D Centers Do. Co. Mo EUROPE 1998. Aug YRP 1998. Mar Do. Co. Mo USA LAB Do. Co. Mo EURO LAB 1999. Nov 2000. Nov Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 7

Do. Co. Mo USA Labs • About 45 employees, 35 researchers • CEO: Dr Minoru Etoh • Charter: Research on future 4 G architectures, services and networks • Since 1999 – IPR: Numerous patents filed and papers published – Substantial presence on standards bodies • IETF: Chairs of 3 IETF working groups, Seat on ISOC and IAB • 3 GPP 2, IEEE 802. XX, etc – Research presence: • Area editors of journals and magazines, Conference TPC memberships and chairs, etc – Collaborations: Several universities, companies and consortia Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 8

Do. Co. Mo USA Labs: Research Themes (1 of 2) • Seamless services over heterogeneous access networks and devices – Ubiquitous services and their control – Browser session mobility, service discovery, context awareness – Seamless AAA • Mobile middleware, platforms and handsets – Content Distribution Networks (CDN): Qo. S, security and APIs – Mobile servlets, lightweight security, disconnected operation • IP-based network architectures and protocols – Fast Mobile IP handoffs and micro-mobility protocols – Location estimation, prediction and privacy – Qo. S estimation, control and applications Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 9

Do. Co. Mo USA Labs: Research Themes (2 of 2) • Multi-hop, ad-hoc and hotspot radio networks – Signaling, Qo. S, cost and capacity – Cross-layer adaptation and WLAN • Security and cryptography – ID and Certificate-based encryption, anonymous credentials, micro-payment schemes and e-commerce • Advanced speech, audio and video coding – Unified speech and audio coder, Mobile 3 D audio and hi-fi video telephony We welcome meaningful research collaborations and partnerships Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 10

And now to shift gears … The Feature Presentation Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 11

Outline: Environmental Design for Pervasive Computing Systems • Do. Co. Mo USA Labs • • • Motivation and background Using less Using it longer Smart disposal Conclusions Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 12

Recycling Council of Ontario Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 13

Frequently Raised Objections (FRO) • Just throw all the old PCs and cellphones into the sea – Ahem … • This is a naïve and ambitious plot to overthrow the capitalist system – No … we are just promoting environmentally sustainable design – Environmental responsibility has been embraced by many major corporations and governments • This is not really a big problem – We hope to show data that indicates otherwise • This is not my problem … why would I (or anyone) pay for this? – Legislation and consumer pressure is building – Market-leader advantages • This is not a computer science problem – We believe software, not hardware or materials, is increasingly important • This is not a research problem – We believe environmental factors can affect all layers of system design – Analogy with battery power considerations in mobile computing Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 14

Computer waste (so far …) • • • Estimate: Over 75% of all computers ever bought in the US are stored in people’s attics, basements, garage … (MCC, 1996) The growth of electrical/electronic waste is 3 times the growth of other municipal waste (AEA, 1997) Estimate: By 2004 there will be over 315 million obsolete computers in the US alone (NSC, 1999) – 4 billion pounds of plastics waste alone • Toxic materials in PCs (MCC, 1996) – – – Lead, Mercury, Cadmium, Arsenic, Chromium Polyvinyl Chloride (PVC) Polybrominated Diphenylethers (PBDE) … “New personal computers release over 100 different chemical compounds as gases, adversely affecting the health and performance of office workers”, • • Technical Univ. of Denmark study, Environment Daily, 9 Sep 2002 Accounts for a significant proportion of U. S. energy consumption – ~3% of residential use, 5 -10% of commercial use, and growing fast (Rosen & Meier, 2000; EPA, 2003) Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 15

This is just the beginning Source: Rainer Malaka, EML ICDE 2001 Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 16

Pervasive Computing • Most pervasive computing devices consume less material than traditional PCs • However pervasive computing devices: – Will be far more numerous • Example: 13 M Bluetooth devices shipped in 2001; Expect 780 M by 2005 (Cahners In-stat) … compare with black phones – Rapid replacement due to low cost or immature technology – Can be disposable • Example: disposable cell phones (Telespree, Hop. On) – Will be embedded in other products • Jewelry, clothing, smart floors, sensor networks, … • Makes location, extraction, collection, recycling harder – More likely to be lost, forgotten, or simply abandoned – Use batteries, likely exclusively – Will bring computer environmental impacts to regions where none exist at present Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 17

A Challenge • Reduce total-lifecycle per capita computer & communications environmental impacts – Material and energy consumption by 10 x – Non-recycled material and non-renewed energy consumption by 100 x – Toxic and harmful byproducts by 1000 x Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 18

Design motivations • Not “The sky is falling …”, but pervasive computing does pose a new environmental risk • Design for the environment – Reduction of resource consumption – Reuse of resources and products – Recycling * Needs to be an integral part of the design process, not an afterthought * Recycling is important, but is not the answer * Software is increasingly important • We have an opportunity to do this while still at the start of the technology wave • Legislative and consumer pressure – Extended Producer Responsibility (EPR) in Europe • First-mover advantage (e. g. NEC Eco PC) – Many “socially responsible” advances are first resisted (e. g. Cellular 911) – Foresight sees market opportunities and differentiators Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 19

One Example: NEC Eco PC • • Oct. 2002: NEC Powermate Eco receives PC Magazine award for technical excellence in desktop PC category Editors cite: – The only desktop PC that doesn’t use a fan • Uses less energy, is quieter – The first to address PC recycling • • • Motherboard uses lead-free solder Chassis is made with 100% recyclable plastic Chassis plastic uses non-toxic flame retardant • The largest opinion expressed by visitors at the NEC Internet Environmental Forum is “Like NEC because of its positive stance” • NEC President K. Nishigaki: “We are coming to a stage where, if a product is not environmentally sound, consumers will not buy it”. Source: NEC 2002 Annual Environmental Report, http: //www. nec. co. jp/eco/en/annual 2002/ Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 20

Using less: Minimizing physical materials • Minimizing physical materials – Doing more with less, not Doing with less • Two parallel trends – Integration: cell phone as pager, organizer, e-wallet, radio, media player etc. . . – Specialization: different functionality, form factors, power requirements, connectivity, processing and storage, fashion niches • Reducing physical materials within a single device – Modular design should allow configuration on a per-user basis – Similar to PC configuration options but with smaller option units • Device sharing – Example: Environmental impacts of answering machines ~10 x more than centralized voicemail (Taiariol, 2001) – But voicemail still lacks features (e. g. live screening) and a good UI Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 21

Using less: Device sharing on a larger scale • 150 M hosts connected to the Internet, mostly underutilized – Have been harnessed for tasks that would be impossible otherwise • Mersenne Prime Search (GIMPS): 130 K users, 1. 5 TFlops • 213466917 -1 (4 M digits), Michael Cameron (Age 20), Canada, 800 MHz PC • Two key challenges (Anderson & Kubiatowicz, 2001) – Internet Scale Operating System (ISOS) for resource allocation, security … – Economic models to provide private owners with incentives • Additional challenge: take environmental concerns into account – Energy: Preferentially utilize computers in cold regions, or those that are not in dormant mode, or depending on available cycles – Materials: Utilize CPU and storage that would otherwise be wasted • Sharing in enterprise and wireless networks – Pervasive computing environment in user’s home, office, car – More device & network heterogeneity, limited resources, restricted UI Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 22

Using less: Minimizing energy usage • Formal models of energy consumption (Big-oh joules) – Needed to motivate algorithm improvements – Analogy to formal models of disk I/O • Energy-efficient applications and architectures – Example: Half-duplex multiparty calls (“Push-to-talk”) use 50% less power than voice calls • Most 2 G/3 G systems do not support this feature • Need to consider total-lifecycle energy impacts – Cellphone consumes ~6 k. J/day. Charger: ~110 k. J/day (Nicolaescu, 2001) – Do. Co. Mo cut total consumption from 115 k. J/day (1992) to 3 k. J/day (2002) – Nearly 90% of energy usage in cordless phones and answering machines is in standby mode (U. S. Do. E, 2002) – Consider energy consumed in manufacture, distribution, and disposal • Alternative energy sources – Solar energy – Human energy: wrist movement (Citizen eco-drive watch), footfalls can generate 50 m. W (Paradiso, 2000), keystrokes Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 23

Using it longer Programmability • Software Sprawl – SLOC = 13992 e 0. 74 yrs R 2 = 0. 98 SLOC = Lines excluding comments and blank lines Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 24

Using it longer Programmability • Software Sprawl Again Office Productivity Suite Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 25

Using it longer Programmability • Most users do not need all the features of the program (or even know they exist) at any one time • Need mechanisms to – – discover system capabilities discover applications and components auto-configure software secure, just-in-time, just-right plug-in upgrades • Hardware requirements should be written not only for entire applications, but based on user-level features • Automated testing techniques need to be developed that support modular hardware • Similar considerations apply to data sprawl Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 26

Using it longer: Extending storage space • Estimate: 30 -60% of disk space on a computer is wasted (Lyman & Varian, 2000) – Example: Multiple revisions of the same document, or outdated software • This Data Sprawl also costs energy (for search and retrieval) and creates usability issues • Need self-managing and self-destroying data • Need better knowledge management support, duplicate prevention, on-line or automatic compression. Examples: – Few word processors offer journaling to store and locate most recent version – Duplicate deletion needs to be done for user’s entire pervasive environment – E-mail attachments are copied to multiple mailboxes instead of being automatically stored in a location and the URL being sent • “Storage is cheap” hides the environmental (and usability) costs of obsolete or useless data Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 27

Using it longer Just-right hardware upgrades • Design hardware and software to support hardware upgrades • Possible example • The OS for PC Y can treat multiple heterogeneous CPUs and disks as single logical units • Upgraded PC cost = 10% for CPU, 10% for disk, and 5% for memory • Techniques for doing this to a user’s pervasive computing devices are required Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 28

Smart disposal • Recent increase in recycling PCs – Manufacturer take-back legislation in US • Cellphone recycling in Japan (Belson, 2002) – Gold: 24 micrograms/phone. Total metal extracted: 21 cents/phone • Crushing devices into their raw materials loses the vast majority of their value • In addition, there are significant health and safety risks of recycling itself … Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 29

This is not a problem … Source: USA Today 2/25/02 Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 30

• Recycling should focus on – Identifying subassemblies and larger blocks for reuse • RF ID tags – “Upcycling” or remanufacturing into new products • Close the loop of product information Information Associated with Product Smart disposal – Provide quantitative Remanufacture feedback to designers on Resale actual use, upgrade and Production Retail Consumer failure of software and Disposal hardware Thomas, 2001 Product Lifecycle Stage Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 31

A process response: ISO 14000 • Standards & guidelines issued in 1996 for environmental management – – • Determining the environmental impact of an organization’s activities Establishing goals and targets for those impacts Evaluating how well the goals and targets are being achieved Continuously improving performance 2001: Over 20 K certified organizations in 98 countries – Certification requires formal assessment by a neutral third party • Does not provide specific targets or methods to achieve them – Processes, not products • Not a guarantee of improved environmental performance, but – A means of demonstrating voluntary corporate commitment – A set of formal methods towards achieving goals – Canadian survey (1999): almost all certified organizations found ISO 14000 beneficial. Improvements cited: environmental performance, cost reduction and competitive advantage • A Proposal: Need to consider the effects of software and hardware/software codesign in ISO 14000 targets Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 32

Summary • Environmental impacts need to be treated as a first-class design constraint • If applied consistently, they will have impacts at all levels of software and systems design … a major challenge to integrate total -lifecycle costs into all design aspects • Lower layers – – • – – – • Seamless integration of old and new hardware and just-right upgrades Environmental parameters to an Internet Scale OS Pervasive OS with environmental design Minimize OS and software sprawl – – • – Reduced requirements Intelligent hardware diagnostics and workarounds Present surviving resources to apps – – • • • Minimize data sprawl with better knowledge management, duplicate avoidance, retrieval, and automatic compression Discover system capabilities, software component discovery and composition, secure, just-in-time plug in Compiler and automated testing techniques Applications and UI – – Low-cost fault-tolerance techniques – – Applications support – Hardware requirements specification based on user-level features Modularly upgradeable hardware Identify subassemblies for recycling Software radios to adapt to new protocols and air interfaces OS and system software – • Just-right software upgrades Energy-efficient applications and architectures Improve UI for device sharing Make energy and toxic byproducts visible to users: • Energy. Star -> Non. Toxic. Star Formal models of energy consumption New design and modeling methodologies Use of alternative energy (human, solar) Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 33

Barriers • “A threat to growth / capitalism / our way of life” – Sustainable and environmentally responsible design is the best way to promote long-term growth – Carrots (economic incentives, secondary markets, broker opportunities) are better than sticks (legislation, consumer disaffection) • No immediate short-term benefits – But see NEC Eco PC, Electric cars, etc • Few professional incentives for R&D in this area – National or international R&D funding programs are needed – No workshops, journals or “tenure-friendly” forums and outlets • Research area is difficult: – need multidisciplinary approach (e. g. air bags) – need precise problem formulations – no established benchmarks Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 34

Summary • Do. Co. Mo USA Labs is doing exciting research on 4 G wireless services, architectures, and networks • Pervasive computing offers exciting opportunities but also possible negative environmental impacts • We need to treat environmental impact as a first-class design constraint • Increasingly software is a key to reducing environmental impact • We need research in many areas of system design for pervasive computing and communications Copyright © 2003 Do. Co. Mo Communications Laboratories USA, Inc. All Rights Reserved. 12/19/2021 Ravi Jain 35