IoT versus M2M; Barriers or Synergies

Visiting two interesting shows co-located at the Portes de Versailles exhibition center in Paris last week on IoT and M2M technologies highlighted, for me, the need to understand clearly what the IoT is.

Discussing with colleagues, I received some interesting remarks that reminded me of the "IoT-versus-M2M" mentality that can persist in the electronics industry. Faced with recent IoT "hype," some people who have long been involved in M2M (Machine-to-Machine) technologies may show signs of incomprehension… even disdain.

This quote does not represent a bad observation. It's just an indication that the definition of the IoT, by comparison with M2M, is not clear... even to people with significant technical background. This got me thinking about what is the real difference between the IoT and M2M, and how we define these technology trends.

Machine-to-Machine: Inter-Connecting Systems

The notion of connecting two machines with an electrical communication channel is very old, dating to the early 20th century by some accounts. But the current sense of M2M, is that of multiple-machines connected to a hub. Data from the connected machines can be analyzed centrally. This centralized analysis can then result in coordinating and triggering actions by any of the connected machines or by humans monitoring those systems.

Perceptions of M2M that I have identified vary from one person to another on the following points: 

? Does M2M imply centralized data collection, or only communication between two systems to coordinate actions?

? Does M2M encompass triggering of human intervention, or only interconnection of machines?

For me, automated production lines are a good example of an M2M environment. Imagine a robot that paints parts on an assembly line. Robots, pumps and conveyors manipulate parts, tools and paint. All of the actions of these systems must be coordinated. This implies coordination between the machines. It also implies monitoring and supervision to trigger actions that are outside of the scope of the machines themselves (for example triggering human interventions such a resupplying paint).

There are de-facto standards for communications in M2M environments, although the concept itself does not dictate a specific type of communication (wireless versus wire) or protocol. Wire communications are the most common for security, robustness and speed of transmission. However, bi-directional communication is essential to triggering of machine actions.

To summarize, M2M clearly includes:

? Electrical systems

? Interconnection (communication, networking)

? Sensing and triggering

? Monitoring (data collection)

? Supervision (data analysis, decision, triggering relevant actions)

M2M is very much a general concept of interconnected systems found in many environments; manufacturing, transportation, infrastructure/building management, etc.

So what is the IoT by comparison? What does it capture, that M2M does not?

Internet-of-Things: Connecting to Things

The “Internet-of-Things” (IoT) is a term and associated concept that was coined by Kevin Ashton. Keep in mind that Ashton has experience with RFID (Radio Frequency IDentification) technology - electronic tags that use magnetic waves to power themselves and communicate with a system. I find this important, because RFID extends the range of data collection from networked machines to networked "objects." RFID goes very much in the sense of Ashton's IoT concept.

As Ashton describes it, IoT is not M2M. And, he is adamant on this point. He sees M2M as the interconnection of systems. He projects, in the IoT, a world in which optimization of computing, sensor and communication technologies make it possible to connect previously unconnected objects (things) to networks. And when we think of objects, consider objects that previously didn't even have electronics in them.

Imagine a bridge allowing passage across another highway. Not a complex, mechanized installation, but an ordinary structure with no associated machinery or electronics. Equip the structure with sensors for precise geo-positioning, vibration and temperature detection. Each sensor is strategically placed to provide data to monitor use, structural shifting and wear. Then, relay these sensors to a supervisor capable of analysing their data and triggering interventions.

Recuperating sensor data, even occasionally can improve inspection efficiency and help reduce human errors in inspection. Recuperating enough data over time can contribute to improvements in the structure and the management of its use. This is the IoT in a simple form.

Important ideas in Ashton's vision include:

? Data from sensor-augmented "things"

? Interconnection (communication, networking)

? Monitoring (data collection)

? Supervision (data analysis, decision, triggering relevant actions)

Key technology evolutions that contribute to Ashton's vision:

? Sensor technologies their miniaturization, precision and robustness

? Reduction of power consumption making sensor deployment in un-powered things possible

? Advances in computing at both the "thing" and the "cloud" levels

? Data analysis techniques that contribute to our understanding of data and to decisions (automated or human)

Recent advances in machine learning can also be considered potential contributors to the IoT. These contribute to greater efficiency in analysis and automation of decision processes.

Note: As we move from MtoM to the IoT implementations, enlarging the scope of our information networks has tremendous security implications. How do we police all the newly connected things to avoid hacking, denial-of-service events, etc.? How can collected data be used? Where are the limits between public and personal data? How do we protect individuals from the harms of data theft? In an M2M context these issues were often treated at a very local level (ex. the production plant and its IT organisation). In the IoT, the scope of data collection and networks is rapidly expanding beyond a single organisation.

Where Does this Leave Us?

The Internet of Things calls on us to apply concepts that we have already implemented in machine environments, but at a much more widespread scale. It also calls on us to augment these implementations with advances in computing, data analysis and machine learning.

Essentially, any discussion of the IoT must encompass:

? Data sources whether systems or sensors

? Power generation / transfer, consumption and optimization

? Computing technologies at the thing and cloud levels... and everywhere in between

? Data analysis techniques

? Machine learning or artificial intelligence

? Security of communication channels, computing and data

? Standards regarding all of these

For many who have been involved in M2M, this may all seem "old hat." This is because the IoT is essentially an extension of what has been done to some extent in M2M environments. This evolution is bringing new uses for these technologies, as well as a host of new challenges. Erecting conceptual boundaries about what “is” or “is not” IoT / M2M creates artificial barriers in discussions about the challenges we face in the very near future.