IoT Turns To Dust

The current thinking for the is that a single or bi-directional interface will be attached to just about everything. It will be an amalgamation of hardware and software that will sense whatever we want, assess it anyway we want, and send it anywhere we want—or where someone else wants.

“There has probably never been a more exciting time to work on semiconductors and their security,” observes Benjamin Jun, CTO and vice president of the Cryptography Research Division at Rambus.

Making everything intelligent and secure is really a massive undertaking that will require an a lot more universality than we currently have, and a lot more time to get it right. Jun notes that, going forward, and this applies to so many platforms, “There is just a lot more than the platform writers, the software creators and the system developers are asking for from the silicon.”

With nanotechnology, that takes on an entirely new dimension, especially in the context of a massive number of connected objects expected on the IoT. Even though a sizeable number of devices that will only have to sense data, without necessarily needing to receive data or take action, these will still require new hardware that is scalable from nano to macro, and securable.

Consider, for example, a flood sensor in a basement. Today, it is still, generally, connected to a central alarm system that will take action if the sensor is activated. There are no two-way communications here because there is no need for it. The job of the sensor, is simply to report a state.

Likewise, that same state report condition will continue to exist in the IoT. The network certainly will be different. It may be connected to an alarm circuit with a Bluetooth radio that interfaces with your network, for example, but not the function – same concept, different interface.

The point is that the sensor itself does nothing but report. It can be as sophisticated as one wants to make it, to report any combination of parameters (temperature, pressure, humidity, wind, light, dark, etc.) to a receiving controller (via wireline, wireless, fiber, etc.) that takes appropriate action, with or without human intervention. However, it still just reports. And as it is now attached to a world-wide network, the electronics and their security take on a whole new dimension.

So, when one thinks about it, sensors could be the majority of IoT devices, especially with the latest concept—IoT dust.

Joy Weiss, CEO of Dust Networks and Linear Technology, says that “there is nothing new about the sensor network. Sensors have been around. A game-changer was the idea to do it at an ultra-low power and without wires. I mean, no wires for both communication and power. That was the evolutionary step that lead to the development of what is now called IoT dust.”

A dusting of sensors
As the article introduction implies, IoT dust sensors are based on nanotechnology. In this case, microeletromechanical systems (MEMS).

These sensors are called “motes,” micro-sensors that can be designed to accomplish a variety of functions. An example of such a mote is shown in Figure 1.

Fig. 1: IoT mote. Courtesy: Harvard Microbiotics Laboratory.

These minute sensors are achieve a high level of sophistication. Micro-scalar architecture integrates an entire system, including an OS (TinyOS – explained later in this article) into a footprint currently just over one centimeter – with half that size not far on the horizon. This is all packed into as small a footprint as possible, and made to last as long as possible. The package contains memory, power (battery, solar cells, energy harvesting) and a wireless transmitter (protocols might be Bluetooth, or perhaps some as-yet-to-be developed wireless platform).

“There is a lot going on in this low-power, ubiquitous sensor segment,” says Ian Morris, principal applications engineer for RF connectivity solutions at NXP Semiconductors. “There are really three key considerations for designing these – communications, security, and power.” Communications, says Morris, is likely wireless, which can be IP, low power Wi-Fi, ZigBee, RFID or NFC. “It has to be secure, something like a secure element, especially if they are sprinkled all over the place. And the important key to all of this is power.”

Adds Morris: “There are some really interesting things going on with energy harvesting also. These devices not only need to have the lowest average energy consumption possible, but another key parameter, is to have very low peak currents, as well.”

Ross Bannatyne, general manager for mass market product line microcontrollers at NXP, says low-power MCUs are finding wide application in the IoT, particularly in the wearables market. “Looking at such devices, you have a wireless interface of some description, an MCU that is connected to a variety of sensor inputs, and the sensors. What is critical here is the ability for the sensor to remain in a low power mode, then wake up instantaneously only when data is received from the sensor. One way to handle this is to do a multi-core MCU that has both a low-power core for sensor data collection, aggregation, and external communications, and a higher-power floating point core for highly computational algorithms.” Bannatyne adds that, “by partitioning the software, you can wake the device up, collect the sensor data at low power, and store it for manipulation later. The point is that you don’t have to turn on the FPU core until you need it to do number crunching or radio usage. And there is quite a difference between the power consumed by the different cores and the less you use the FPU core the lower the overall power budget is.”

What to do with that tiny mote
The applications for IoT dust are limitless, but there are some rather special applications that skirt the edge of the envelope. Focus for a moment on the device in Figure 1. Such a device would be ideal for deployment during a natural disaster where the communications infrastructure is knocked out. Or, in a combat situation where dust eyes and ears can relay enemy data for troop and support ordinance. On a smaller scale, these motes can be deployed within a house, or on a ranch or farm, or even in a forest to report almost any type of event.

What all of these scenarios have in common is that motes would be of the airborne variety, which will be the focus of the rest of this article. Look for other dust articles to appear that will address other types of dust, as this avenue progresses.

Airborne motes have different challenges that hardwired (IP-based) or fixed motes, which can interface to the IoT in a standard fashion (IP). First of all, the deployment is fluid and where they are deployed may or may not have an existing wireless infrastructure. Once deployed, they will require an interface to the network. That may be ad hoc, as well. Another challenge is power. There is no backhaul to a power system, and unlike fixed motes they cannot handle a bulky power pack.

But more than anything, these motes must be able to secure sensitive data such as keys or critical information. “There is a lot of talk about securing the connection,” says Jun, which is a primary concern with wireless sensors. “But overall, the issue isn’t necessarily securing the connection. The real issue is to ensure that the parties on the connection have established keys, credentials, and identity.”

In short, a secure identity can indemnify that senders and recipients are who they are supposed to be.

Securing the motes
As with all IoT devices, securing motes will be a top priority. And, because this is an emerging platform, the industry has as chance to do it right from the start, especially with airborne motes.

“With smart dust, typically, one is not going to go to an IT guy for installation,” notes Rambus’ Jun. “It will likely be sprayed from a helicopter and there will be no enrolment step as part of that process. Therefore, we want to put identity [and security] into the smart dust very early on.”

NXP’s Morris adds, “One of the challenges of securing such devices is that hackers have physical access to them. There is a lot of discussion about securing communications. But if either end can be attacked, security becomes even more vital in devices that are physically deployed and have easy access to them – especially if keys and such are involved.”

That means that such motes will have to hit the air fully functional, and fully secured. They must be able to immediately interface with the network using some sort of secure protocol and retain that security level during reporting.

Mobile motes have a different model than fixed ones. Their interface will almost surely be RF, and they will have to contain some sort of “self-destruct” mode. The latter has more relevance in areas such as military. If, say, a bucket of motes is deployed in a hostile area, such as a rough mountainous terrain in Afghanistan, it could be difficult to insure some won’t fall into enemy hands. If they do, they cannot reveal keys or allow the enemy to hack into the network.

Typical methodologies for securing wireless motes include spread spectrum techniques, standard 128-bit AES encryption algorithms, data integrity via message integrity codes, replay counters, and denial of service (DoS) techniques and algorithms.

The control
One of the more interesting subjects about motes is the operating system. Motes use what is call the “TinyOS.”

TinyOS, like so many open-source projects, was part of the DARPA nest program, and started life as a collaboration between the University of California, Berkeley, Intel Research, and Crossbow Technology. Today it is a full-blown, international consortium called TinyOS Alliance.

It was developed for wireless sensor networks (WSNs) and especially smart dust devices and networks. There are others in this arena, Nano-RK, LiteOS, Contiki, Arduino, etc. with comparable features but TinyOS was chosen because of its robust code, capabilities in networking, low-power operations and abstraction support (for wireless networking).

TinyOS is a C language offshoot, called nestC. It is specifically written for space-constrained applications like dust devices. It has a unique programming interface that consists of non-blocking APIs that take up little resources (RAM). Briefly, how that works because limited resources means it can’t wait for the message to run. It just assumes it does, because there are not supposed to be any long-running codes. Thus the send function returns almost immediately.

TinyOS programs must be written concisely because long code will hang the OS and not allow it to run other calls. The way to get around this is to break up long code into small parts, or snippets, and not have nested loops. The goal is to take the computations and break them into N computations over N cells, as opposed to an NxN array. TinyOS is designed for that.

For this OS, if the code constraints are adhered to, it is an ideal OS for IoT motes, both mobile and fixed.

Conclusion
The IoT is still mostly a moving target. Sources estimate that there are presently about 9 billion devices on the 2014 IoT, most of which are not autonomous. By 2020, however, that will be exactly the reverse. Most the devices will be autonomous, and sensors will be prolific among them.

“Device manufactures are adding more and more sensors into their products across all industries; vehicles, buildings, office, medical, industrials of all kinds, wearables, even ‘invisibles’ — sensors that you don’t know are there,” says Bannatyne. “Sensors are approaching cost levels where, eventually, you can attach them to just about everything, and monitor just about anything.”

There is a lot going on in the areas of sensors, especially in the communications and power segments. This article has mostly been an introductory piece that delved into these segments. Look for future articles that explain these devices and methodologies in more detail.