INVITEDPAPERSoC Issues for RF Smart DustWireless sensor nodes, each a self-powered system performing sensing,communication, and computation, form reliable mesh networkscoordinating efforts to add intelligence to the environment.ByBen W. Cook, Student Member IEEE, Steven Lanzisera, Student Member IEEE,andKristofer S. J. PisterABSTRACT|Wireless sensor nodes are autonomous devicesincorporating sensing, power, computation, and communica-tion into one system. Applications for large scale n etworks ofthese nodes are presented in the context of their impact on thehardware design. The demand for low unit cost and multiyearlifetimes, combined with progress in CMOS and MEMS proces-sing, are driving development of SoC solutions for sensor nodesat the cubic centimeter scale with a minimum number of off-chip components. Here, the feasibility of a complete, cubicmillimet er scale, single-c hip sensor node is explored byexamining practical limits on process integration and energeticcost of short-range RF communication. Autonomous cubicmillimeter nodes appear within reach, but process complexityand substantial sacrifices in performance involved with a truesingle-chip solution establish a tradeoff between integrationand assembly.KEYWORDS|Low-power circuits; low-power RF; Smart Dust;wireless mesh networks; wireless sensor networks; wirelesssensorsI. INTRODUCTION AND HISTORYThe term BSmart Dust[ hascometobeusedtodescribeawide range of wireless sensor network hardware at a smallscale down to a handful of cubic millimeters [1]. Eachwireless sensor node, or Bmote,[ contains one or moresensors, hardware for computation and communication,andapowersupply(Fig.1).Motesareassumedtobeautonomous, programmable, and able to participate inmultihop mesh communication.The genesis of Smart Dust was a workshop at RAND in1992 in which a group of academics, military personnel,and futurists were chartered to explore how technologyrevolutions would change the battlefield of 2025 [2]. Bythis time it was clear that MEMS technology was going torevolutionize low-cost, low-power sensing. Moore’s lawwas accurately predicting CMOS digital circuit perfor-mance improvements with no end in sight, and thewireless communication revolution, already firmly estab-lished in two-way pagers, was beginning to make its wayinto handheld cellphones. The confluence of these threetechnological revolutions in sensing, computation, andwireless communication placed the major sensor motefunctions on asymptotic curves down to zero size, power,and cost over time. Furthermore, the potential forcointegration of CMOS and MEMS made single-chipsensors with integrated signal conditioning possible at lowcost [3]–[11].In 1996, the term BSmart Dust[ was coined to describethe ultimate impact of scaling and process integration onthe size of an autonomous wireless sensor [12]. SeveralDARPA-sponsored workshops in the mid-1990s fleshedout some of the implementation and application details ofthe 1992 vision, and key research proposals were writtenand funded at the University of California, Los Angeles(UCLA); the University of California, Berkeley; and theUniversity of Michigan, Ann Arbor. It was clear to thecommunity at that time that low-cost ubiquitous wirelesssensor networks would have a revolutionary impact onmilitary conflict. What was not as clearly anticipated wasthe potential impact on commercial and industrialapplications.The first wireless sensor motes, called COTS(commercial-off-the-shelf) Dust, were built early in theSmart Dust project using printed circuit boards and off-the-shelf components. It was shown that these inch-scaledevices could perform many of the functions predicted inthe 1992 workshop, including multihop message passingand mote localization [13]. COTS dust and other macro-scale motes were developed to explore sensor networksoftware and individual mote architecture as well as deploysmall scale networks [14]–[16].Manuscript received August 24, 2005; revised February 21, 2006.The authors are with the University of California, Berkeley, CA 94720-1774 USA (e-mail:[email protected]; [email protected]; [email protected]).Digital Object Identifier: 10.1109/JPROC.2006.873620Vol. 94, No. 6, June 2006 | Proceedings of the IEEE 11770018-9219/$20.00Ó2006 IEEEWhile great strides were made in miniaturization andpower reduction of the hardware, perhaps the most im-portant event during this early period was the observationthat networks of autonomous sensor motes represented aubiquitous, embedded computing platform [17]–[20], andthey needed a new operating system to match. Proposed inthe Endeavour project [21], the TinyOS operating system[22] was developed under DARPA funding and put intothe public domain, along with all of the COTS Dusthardware designs, and a thriving open-source sensor net-working community was born.Meanwhile, in 1999 the IEEE formed the 802 workinggroup 15, with a charter to develop standards for wirelesspersonal area networking (WPAN), from which the low-rate WPAN 802.15.4 standard emerged. The 802.15.4standard was designed from the beginning to be a low-power, low-complexity solution for sensor networking inindustrial, automotive, and agricultural applications [23].A spinoff group from the industrial consortium HomeRF,focused on home automation applications, created theZigbee standard in 2004. Zigbee 1.0 [24] is based onthe 802.15.4 standard radio [25]. With the blessings ofthe IEEE on a radio standard, a consortium of largecompanies defining applications, and the help of theventure capital community, a new industry was born.II. DEVELOPMENTS IN SENSORMOTE HARDWAREThe Mica mote (Fig. 2), the most popular mote used inresearch, was developed to mimic the expected architectureof a highly integrated mote while using off-the-shelf partsmounted on a common PC board to reduce developmenttime. This mote includes a microcontroller, RF transceiv-er, and the ability to interface to a variety of sensors. Themote is powered by a pair of AA batteries, and these takeup the majority of the unit’s volume [14]. Similar inch-scale motes utilizing primarily off-the-shelf componentsare now commercially available from numerous sources[26]–[30].Development of highly integrated sensor mote compo-nents started in the mid-1990s and resulted in multichipsystems that could be assembled to create a mote. AtUCLA, MEMS devices were combined with commercialCMOS chips that provided sensor control and readout aswell as