IN THE PIPELINE: 'Smart Dust' Finds A Way Out Of The Lab
Imagine automobile tires that sound a warning when they need air, a milk carton that tells grocers its contents are spoiled, or sprinklers that know when crops are parched.
It may sound like the stuff of a science-fiction movie. But university scientists have created tiny, sophisticated sensors that promise to do these things and much more. One of the pioneers of this research has started a company to bring it to market.
The micro machines, commonly called motes, are able to measure air pressure, temperature and humidity, among other things. They are inexpensive and disposable, yet capable of monitoring their surroundings for several years on miniature battery power.
Motes contain microprocessors, two-way radios and software that makes them "smart" enough to form a wireless network. Scatter a bunch of motes in a field and they are able to locate one another, collect data and communicate with a base station. If one mote fails, the others work around it.
"They are self-organizing and self-healing. It's really revolutionary technology," but it has been slow to find its way out of the lab, said Kris Pister , a professor at University of California-Berkeley who has studied motes since 1998. "It takes a while for people to get their heads wrapped around it."
Earlier this year, Pister took a two-year sabbatical from Berkeley to launch Dust Inc. with two of his former Ph.D. students. He has raised about $1 million from wealthy individuals, including Microsoft Corp. (MSFT) scientist Gordon Bell, and is seeking more funds.
Dust plans to refine the hardware and software created at Berkeley and sell it to sensor manufacturers. "Our goal is to be the premier supplier of the chips and operating system that makes wireless sensors work," the 39-year-old says.
There have been four generations of motes developed since Pister began his research five years ago. Early versions were about the size of a small mobile phone, ran on AA batteries and cost about $100 apiece. Newer ones are as small as a soda bottle cap.
Dust's mission is to shrink the size and cost of these tiny sensors by combining multiple components into a single tiny chip. In March, the company demonstrated a prototype measuring 2 millimeters by 2.5 millimeters - small enough to fit on Lincoln's face on a penny.
The chip had to be connected to an antenna and battery, but Pister says the goal is a "fully packaged mote" that's roughly the size of an aspirin. He thinks such devices could cost as little as $1 within five years.
Intramural Competition
Dust is not alone. There are several companies working on tiny sensors and two-way wireless networks in what has become an intramural competition. Among them is another startup with ties to Berkeley, Crossbow Technology Inc., and two companies founded by researchers from the Massachusetts Institute of Technology.
"I'm shocked at the way the technology has advanced," says Robert Poor, who researched wireless sensor networks at MIT before starting Ember Corp. in 2001. It's going to continue to improve and "get less and less expensive."
The Boston startup, whose directors include 3Com Corp. (COMS) founder Bob Metcalfe, no longer builds sensors. Instead, it uses tiny radio chips from a Norwegian partner, Chipcon AS, and sells the necessary networking software and tools to sensor makers. Ember says its software and a 9 mm by 5 mm radio chip cost as little as $5 in large quantities and last up to five years on battery power.
"We've got a really healthy pipeline of customers in various stages of development," said Poor, Ember's chief technology officer. Among them is Tyco Thermal Controls, a unit of Tyco Ltd. (TYC), which has tested using Ember's technology in temperature sensors at oil refineries.
The other sensor company spawned by MIT is Millennial Net Inc. of Cambridge, Mass. It was founded in 2000 by Sokwoo Rhee, who designed wireless biomedical sensors as a doctoral student at the university. Millennial Net's devices, which it calls i-Beans, measure 28 mm by 23 mm (or about 1 square inch), can transmit up to 30 meters and run on coin-sized batteries for up to five years. These miniature computers are connected to, or placed inside, existing sensors to create "smart" wireless networks.
One of the startup's target markets is remote health monitoring: wireless devices that regularly test glucose levels, pulse or oxygen saturation, "get that information off your body and present that data back to you or your physician," said Tod Riedel, Millennial's president. Riedel thinks i-Beans will ultimately cost less than $10 apiece.
"Wireless is hot for the sensor community," said Jack Sun, chief executive of Helicomm Inc., a Carlsbad, Calif., company that sells wireless sensor gear and software. The subject dominates industry conferences and continues to attract venture capital, he adds.
Nonetheless, one high-profile sensor startup, Graviton Inc., shuttered its doors earlier this year. The La Jolla, Calif., company, founded in 1998, set out to dominate the market by doing everything from research on tiny sensor components to running a network operations center for customers.
It hired seasoned executives, inked partnerships with Motorola Inc. (MOT) and Qualcomm Inc. (QCOM), and reportedly raised more than $60 million from blue chip investors, including Kleiner Perkins Caufield & Byers and the venture arm of the Central Intelligence Agency. But in the end its appetite proved to be its undoing. Last month, Graviton's assets and intellectual property were acquired by Xsilogy Inc., a San Diego sensor startup, for an undisclosed amount.
"Graviton, consistent with how funding was done then, had a charter to do it all," says John Major, a former Graviton board member and industry veteran. "It did a lot, but when funding ran out it didn't have the time to scale back."
Graviton is a sober reminder of pitfalls that face young technology companies, but Pister is undeterred. "There is some benefit to people having seen the excitement of the future and being let down by hype," he said. "Having stuff that works makes a big difference."
'Smart Dust'
Pister's early research was funded by the U.S. Defense Advanced Research Projects Agency and came to be known as the "Smart Dust" project. The Pentagon wanted to scatter tiny wireless sensors on a battlefield to secretly monitor enemy movements. It was part of a larger collaboration involving Berkeley and Intel Corp. (INTC).
The sensors were so small - they usually have a few thousand bytes of memory and 8-bit processors - that the researchers had to create an operating system that didn't consume much memory or power. They called it TinyOS.
Berkeley made the TinyOS software freely available and shared the blueprints to its sensor hardware. It also partnered with Crossbow, a startup founded by a Berkeley graduate student and the dean of its engineering school, to manufacture the prototypes.
The San Jose company, whose financial backers include Intel, sells mote development kits for between $900 and $2,000 to university labs and corporate research teams. More than 200 groups are working on these tiny sensors, said Mike Horton, Crossbow's founder and chief executive.
"We have found a lot of success in solving everyday problems," Horton said. The tiny sensors are being tested in a variety of commercial uses, including environmental monitoring and asset tracking.
Honeywell International Inc. (HON) has examined sensors from Dust and its peers for use in its security systems and building controls. Among the issues being studied are sensor battery life and managing the data that the sensors transmit, said Dan Sheflin, chief technology officer at Honeywell's Automation & Control Solutions division. He says, "we are looking at moving from pure technology to reality."
“智能微尘”有望投入应用
让我们想象一下，汽车的轮胎在需要补气时会发出警告，奶箱告诉杂货店老板它里面的牛奶已经变质，浇灌机知道什么时候庄稼乾旱缺水。
这听起来象一部科幻影片中的情节。但是，大学的科研人员已经发明了一些轻巧的、先进的传感器，甚至能做比这更多的事情。一位研究人员率先创立了一家公司，希望将这类传感器推向市场。
这种普遍称之为“微尘”(motes)的装置还能够测量气压、温度和湿度等。这种一次性产品成本低廉，可以同它来监控周围环境，仅依靠微型电池就能工作多年。
微尘包括微处理器、双向无线电接收装置和使它们能够组成一个无线网络的软件。将一些微尘散放在一个场地中，它们能够相互定位，收集数据并向基站传递信息。如果一个微尘功能失常，其他微尘会对其进行修复.。
自1998年以来一直从事此项研究的加州伯克利大学(University of California-Berkeley)教授克里斯?皮斯特(Kris Pister)称，它们能够进行自我组织和纠正，这的确是一项创新的技术，但是将其推广应用的过程十分缓慢。
今年早些时候，皮斯特向学校告假两年，和他以前的两位博士生成立了Dust Inc.。他向微软(Microsoft Corp., MSFT)的科学家戈登?贝尔(Gordon Bell)等人士筹资了约100万美元，并且还在寻求更多资金。
Dust计划对学校里发明的硬件和软件加以改进，将其销售给传感器生产商。皮斯特称，他们的目标是成为无线传感器赖以运行的晶片和操作系统的首要供应商。
自从皮斯特在5年前开始此项研究以来，已经开发出了4代微尘。早期的发明和体积较小的手机一样大小，使用AA电池，成本约100美元一个。新的发明只相当于汽水瓶盖的尺寸。
Dust的任务是通过将多个元件整合到一个单晶片中，将这些微型传感器的尺寸和成本进一步缩小。3月份，公司展示了2×2.5毫米的样本，和美国分币上林肯的脸差不多大。
这种晶片需要和一个天线和电池连接，但皮斯特表示，它们研究的目标是把天线、电池和晶片整合到一起，做成一个阿斯匹林药片大小的微尘，估计这样一个微尘的成本在5年内会降至1美元。
Dust并非独家开展此项研究的公司，还有许多公司在进行小传感器和双向无线网络的开发。院校内部及院校之间展开了竞争。另一家初创公司Crossbow Technology Inc.也和伯克利大学有关，还有两家同类公司是由麻省理工大学(Massachusetts Institute of Technology.)的研究人员创立的。
Ember Corp.的创始人、麻省理工大学的罗伯特?普尔(Robert Poor)称，他对此项技术的突飞猛进大为震惊。他认为相关技术将继续得到改进，而且成本越来越低。
Ember不再生产传感器，而是使用挪威的合作夥伴Chipcon AS的小型无线电晶片，向传感器生产商出售必要的网络软件和工具。Ember称，其软件和9×5毫米的无线电晶片大批量生产的造价低至5美元，靠电池可以持续运行5年之久。
Ember首席技术长普尔说，公司在各个研发阶段都获得了良好的客户群。泰科(Tyco Ltd., TYC)旗下的Tyco Thermal Controls在炼油厂使用的温度传感器就运用了Ember的技术。
麻省理工大学创立的另一家公司是Millennial Net Inc。该公司于2000年由Sokwoo Rhee创建。Millennial Net的i-Beans装置尺寸为28×23毫米，发射的信号范围最长达到30米，硬币大小的电池可用5年。这些装置可以连接到或放置在现有传感器内，形成精密的无线网络。
远程健康监控是该公司的目标市场之一。Millennial的总裁托德?里德尔(Tod Riedel)说，通过这种无线装置，可以定期检测人体内的葡萄糖水平、脉搏或含氧饱和度，将信息反馈给本人或你的医生。里德尔认为，i-Beans最终的价钱将不到10 美元。
Helicomm Inc的首席执行长杰克?孙(Jack Sun)称，无线网络在当今传感器行业炙手可热，各个行业会议都热衷于讨论这一主题，它将继续吸引风险投资。
但是，一家拥有较高声望的传感器初创公司Graviton Inc.今年早些时候倒闭。该公司始建于1998年，业务从微型传感器元件的研究到运行客户网络操作中心无所不包。
它聘用了经验丰富的管理人员，和摩托罗拉(Motorola Inc., MOT)、高通(Qualcomm Inc., QCOM)达成合作协议，据说从蓝筹股公司筹资6000多万美元。但是最终它的野心未能实现。上月，Graviton的资产和知识产权被Xsilogy Inc.收购，收购金额不详。
Graviton前任董事会成员约翰?梅杰(John Major)是这个行业的资深人士。他说，Graviton的摊子铺得太大，钱用完了，想要缩小规模已经来不及。
Graviton的这一事件昭示了新兴科技公司面临的潜在隐患，但是皮斯特并未因此动摇。
皮斯特初期的研究是由美国的国防高级研究计划局(Defense Advanced Research Projects Agency)资助的，该项目命名为“智能微尘”(Smart Dust)。美国五角大楼希望在战场上放置这种微小的无线传感器，以秘密监视敌军的行踪。
这些传感器十分微小，研究人员不得不建立一种无需消耗太多内存或电力的操作系统。他们把这种系统取名为TinyOS。
伯克利大学将TinyOS软件免费供公众使用，它还和该校一毕业生及工程学院院长创立的公司Crossbow合作生产样品。
Crossbow的创始人、首席执行长麦克?霍顿(Mike Horton)表示，公司以900-2,000美元的价格向大学实验室和公司研究部门销售微尘开发工具包。他说，他们的产品在处理日常问题上取得了很大成功，并且正在各种商业用途上进行测试，包括环境监测和资产追踪。
霍尼韦尔(Honeywell International Inc., HON)在其安全系统和控制方面检测了Dust及其同行的传感器。霍尼韦尔自动化和控制(Automation & Control Solutions)业务的首席技术长达恩?谢福林(Dan Sheflin)说，他们正在探讨如何使这种技术得到实际应用。正在研究的问题包括传感器电池的寿命和如何管理传感器传递的数据。
