Leading technology cheerleaders, including Ethernet co-inventor Bob Metcalfe and networking siblings Judy and Deborah Estrin, have been touting the potential of distributed sensor networks for close to a decade. In fact, the field attracted the attention of venture capital in the early 2000s, with ambitious startups such as Dust Networks Inc. receiving several million dollars to build out distributed networks capable of monitoring a given area to compile data in realms ranging from smart metering to environmental air quality.
The original vision assumed the existence of microcontrollers cheap and powerful enough to deploy as heavily as the smart dust DARPA talked about when Dust Networks was launched. Sensor networks could be as localized as a single human body (or even a microscopic region) or as widespread as a tropical rain forest or the boundaries of an ocean.
Sensor-net developers could leverage the low-cost wireless technologies designed for RFID and Near-Field Communications markets, but there still were practical limits to the amount of data that could be recorded and classified in real-time.
In the last couple of years, that hurdle has been tackled, if not conquered, by very low-cost microcontrollers following common instruction sets such as ARM, and by centralized data-acquisition nodes that can occupy a space the size of a PCMCIA mezzanine card. This advantage has allowed everything from a hummingbird-sized drone to a dongle add-on to a smartphone to become capable of acquiring, recording, and storing real-time, high-resolution image data. The tools are there.
Proponents of constant real-time data acquisition for health care or environmental remediation should welcome this, while privacy advocates should recognize just how detailed a continuous examination of the environment is possible. But neither side should assume that low costs of components and high-performance, real-time capabilities mean sensor nets will be deployed everywhere in a matter of months.
A series of market research reports issued in early August suggest that a sensor-net market of $2 billion a year or greater may be tantalizingly close, but that its realization still is far from assured.
IDTechEx of Cambridge, England, predicted that the market's size, currently $450 million a year, could exceed $2 billion in 2021, not counting peripheral markets such as location services, which could add $1.5 billion or more to the total. This assumes, however, that first-generation mesh networks used in process control and smart metering are augmented with larger nonmesh networks for forest fire, earthquake, and disaster alerting.
IDTechEx CEO Raghu Das says this assumes some local and national government funding to augment private deployment of sensor nets. He also says the confusing mix of protocols, dominated by ZigBee but also including RFID, low-energy Bluetooth, WirelessHART, and full TCP/IP, can hinder acceptance.
ABI Research sees particular opportunities for the Body Area Sensor Net, where as many as 400 million Wearable Wireless Sensors could be deployed by 2014. The applications are not limited to hospital or clinic uses. They also include fitness and wellness monitoring, where vendors such as Freescale Semiconductor anticipate a bigger market than the hospital BASN application.
But Kirsten West, principal analyst at West Technology Research Solutions LLC, cautions that there are plenty of near-term problems in even the simplest smart meter deployments, and that the problems are the worst in the US.
A WTRS third-quarter report on wireless sensor networks cautions that many network sensor chips deployed in utility company applications have not been activated, and that many subsystems shipped without a software stack. If the scenario in smart metering is duplicated in other markets, sensor nets may be rolled out in prototypes but not used in the actual collection and aggregation of data.
West says the best near-term opportunities for sensor nets in most vertical markets remain outside the US.