Top technology blog Read/Write Web has been one of the most proactive and prolific sources of information about “the Internet of Things”, which (broadly defined) describes the increasing proliferation of network-connected devices and sensors.
Today, editor Richard MacManus published a piece titled “Three Sensor Data Platforms to Watch.” The post (and the rest of R/WW’s coverage in this area) is recommended reading for any technology enthusiast or industry watcher.
(Update 9/14/10: GigaOM adds to the conversation in "Sensor Networks Top Social Networks for Big Data")
Richard rightly notes that much work in the space is “experimentation”, adding, “it’s still very early in this area.” However, HP’s collaboration with Shell is worth bringing to the discussion because it is a particularly large-scale commercialization of the broad portfolio of technologies required to bring this vision to life.
If you’re unfamiliar with this collaboration, here’s a quick FAQ and some video to get you up to speed:
What are HP and Shell doing together?
HP and Shell will use their complementary knowledge and experience to produce a groundbreaking solution to sense, collect and store geophysical data. The solution promises to help the oil company avoid the environmental impact of drilling unnecessary wells thanks to previously unattainble high-resolution data gleaned from a wireless seismic imaging network.
What’s unique about the technology?
The Shell system underscores the uniqueness and importance of HP’s broad portfolio. The new sensing technology represents a breakthrough in nano-sensing research and uses the fluidic MEMS technology codeveloped over the past 25 years by HP Labs - the company’s central research arm - and the company’s Imaging and Print Group.
It will be delivered by HP Enterprise Services and uses HP ProCurve networking products along with HP storage, computation and software products.
The system also encompasses a one-million node wireless network.
What does this collaboration mean for future sensor network applications?
Dr. Peter Hartwell, HP Labs (excerpted from the video below):
“…we’ve been able to use [this opportunity] to kind of pull the technology out of the lab, get involved with our business units, create this whole new system around CeNSE, which, as we develop this and you push things forward, and you work on the cost and size and power requirements, we’ll be able to…get towards actually doing a trillion sensors a lot of applications that are much more important to measuring our impact on the Earth.”
Why hasn’t this been done before?
HP Senior Fellow and nanotechnology pioneer Stan Williams (excerpted from the video below):
“The problem is that in this area there are no standards…when you get a bunch of [suppliers] in the room talking about the system, first of all everyone’s worried about intellectual property. They don’t want to share. So the fact of the matter is when somebody tries to bolt a system like that together it turns out to be a pretty unworkable Frankenstein’s monster.”
“Whereas at HP, we’re actually doing this soup to nuts. We’re vertically integrated throughout the entire stack in terms of being able to deliver this solution.”
What about privacy and security?
Read this post about HP's research into the privacy and security challenges of sensor networks.
Where can I get more information?
Try this post focused on HP Labs accelerometer sensor or read the presentation below
Sensor networks like CeNSE are enormously complex and can cover huge geographic areas. They also span a wide range of applications that can require radically different overall design solutions.
That means that different CeNSE-type networks will also face very different issues of security and privacy, says Bill Horne, a Princeton-based research manager in HP’s Systems Security Lab.
“A network, for example, that consists of stakes that get stuffed in the ground with mini computers sitting on top,” he says, “will face very different vulnerabilities from one where you have microscopic, mobile sensors that get spread out over a big, constantly shifting area.”
However , says Horne, the same process can be applied to addressing the vulnerabilities of each.
“What you need to do is build a threat model,” Horne explains. These are designed to address security concerns from three main perspectives:
1. Confidentiality – to make sure data is kept private
2. Integrity – to make sure no one can change or corrupt the data being collected
3. Availability – to make sure people can't launch attacks (such as Denial of Service attacks) that bring down whatever is important to make the system run.
A threat model also imagines who might want to inflict damage upon a system. Such people could be competitors, angry employees, or locals wanting to steal devices for parts.
But in many cases, says Horne, “your most troublesome adversary will be nature. If you're creating a network out in the middle of the desert, say, and it's 140 degrees and you've got animals running around and wind and sand storms, those are going to be the main threats to the integrity of your system.”
New technological challenges
Running a threat model will often identify issues that can be addressed with existing solutions. But Horne expects CeNSE networks to expose a number of new research challenges, too.
“One issue that has already popped up,” he says, “is low-powered cryptography.”
Conserving battery power is a huge concern with remotely-located sensors. And yet the data they transmit will often need to be encrypted for security reasons, which is typically a computationally intensive activity. “So whether you can design new kinds of crypto algorithms that are power sensitive,” says Horne, “is a whole interesting research field.”
The intersection of privacy and security
While CeNSE-type networks need their own integrity, they also raise issues of privacy for people who are moving about within their orbit.
“Privacy is essentially another confidentiality issue,” says Horne, “only what you care about here is the confidentiality of a third party.”
Such concerns can be identified through building a threat model and to some degree mitigated by good system design. But in the end, says Horne, “it's largely a policy issue; in the sense that people need to know that they have some rights relating to their data and to understanding how it is being used and how it's being shared with other people.”
The OECD, he notes, has established Privacy Principles that define the rights citizens should have to their own data. From HP's perspective , says Horne, “for any application, we would take a very close look at how we're handing personally identifiable information and make sure that we're doing the responsible thing with that information.”
[Editor's note: for more on HP's research into privacy and cloud computing, read this post about EnCoRe]
Putting it all together
Addressing issues of security and privacy in any complex system is hard, adds Horne, and easy to get it wrong.
“Sensor networks are not just a single set of computers,” he notes. “It’s the sensor network, the communication network, the data centers, the people involved. And then all the different stakeholders that come into play each bring different security problems to the table.”
But HP has a centralized privacy organization, he notes, and a services arm with an established expertise in addressing security across the IT landscape.
“It's a complex problem,” says Horne, “and you need a company with the breadth of expertise that you find at HP to do it right.”
HP’s vision for a Central Nervous System for the Earth (CeNSE) is leading to a new generation of computer networks that are aware of the environment in which they operate.
Those systems promise to help us better understand how we are affecting the planet, and to suggest specific actions that we can take to live more sustainably.
But how do you get these networks to actually be aware: to taste, touch, smell, see and hear?
“You’ve got to turn those notions into quantities you can measure,” answers Peter Hartwell, leader of the digital microelectromechanical systems (MEMS) team in HP’s Information and Quantum Systems Lab.
We already have cheap but high-quality sight and sound sensors, Hartwell points out. “Just pull out a modern cell phone and it's got a good microphone and most have multi-megapixel cameras on the back.”
Sensors that can record touch – called accelerometers - have been slower to develop, though. They are being used in consumer applications like vehicle airbags, game controllers, and smartphones. But Hartwell and colleagues have created a groundbreaking new accelerometer (which detects both motion and vibration) that is around 1,000 times more sensitive than the typical ones on the market.
This new sensor is being deployed in HP’s first real-world application of CeNSE – a collaboration with Royal Dutch Shell that promises to help the oil company avoid the environmental impact of drilling unnecessary wells thanks to data it can glean from a high-resolution seismic imaging network.
MEMS for that application must be rugged enough to work in remote locations. They also have to be packaged with a wireless radio, a battery and a solar cell to give it power. Plus they need to be not much larger than a pushpin in size. “It’s a challenging integration problem," says Hartwell.
But it’s also one HP is highly familiar with – millions of MEMS are already at work in its inkjet printer cartridges, which are similarly complex and sturdily built. “In both cases,” Hartwell notes, “you have a complex chip that must be exposed to the environment—to measure it or to squirt ink onto it—and packaged into an integrated unit.”
HP’s experience with inkjets gives it a huge head start, says Jonathan Eunice, an analyst with Illuminata. Expertise in miniaturization, he suggests, “has enormous impact on what you can sense, and whether you can afford to sense things.”
The cheaper sensors get, the more you can afford to deploy, which means you can take more measurements, says Eunice, “and basically when you’re talking about science or engineering, more measurements, more data points, leads to better results.”
New sensors that can ‘taste’ and ‘smell’ better are Hartwell’s next targets. Sensitivity to chemical and biological changes in the environment expands the potential of CeNSE enormously: from bomb-sniffing luggage trackers, to systems that detect pathogens in food, to state-wide early warning networks that can measure the spread of toxins through the air.