IoT as a component of PaaS comprised of three distinct layers…

I was thinking yesterday about cloud, the IoT and of course my three favorite concepts (bandwidth, integration and data). First off the reality of IoT is that it for the most part requires a cloud infrastructure. But like cloud there are components of the IoT that create different solution sets.

For example in the reality of IaaS, PaaS and SaaS, each for the most part relies on the other, except for IaaS. PaaS and SaaS are linked to having a stable and reasonable IaaS infrastructure both the hardware and software of that environment.

Cloud and the IaaS baseline represents a great starting point for the IoT world as well. It makes me wonder if in fact IoT is PaaS v2. Where IaaS is about building and offering infrastructure PaaS is about integration and connection. IoT is simply integration and connection carried beyond human controlled objects to autonomous machines. A good video surveillance system works with or without human intervention.

Within that PaaS environment we should be able to establish some interconnection and integration rules for the IoT that are logical. In effect an IoT sensor is similar to the old three tiered applications we build 15 years ago. A transmission layer, A data layer and a presentation layer. The n-tier environment in which the IoT objects are deployed then represents the place where we can create the interconnection and integration frameworks.

First off we would need to establish the requirements of the overall solution. Not all data needs to be integrated and connected. Some data can simply go to the presentation layer and be consumed by the user.  Weather data for the most part is consume real time data. You don’t often sit down on a Sunday afternoon and review your home weather station data from the past week.  Video data on the other hand you might.  Most video surveillance systems have motion detection options. So you can review any event created by motion related to your system. It may still be a lot (people walking their dogs, kids on the way to school) but it is fewer events and you can sort through 24 hours fairly fast. But you would need those files to exist in your system.

In both the cases stated above we wouldn’t need to integrate with anything. Simply have a system that would support local video DVR capabilities and a way to access the DVR. There are however solutions that would benefit from data analytics. The first would be the actual “atmosphere” of your home. Over time CO2 levels rise and decline in your home. When the levels are high you get a headache or worse. But why are the levels high? When are the level’s high? That is information that you need to be able to collect (via a sensor installed in your home) and then over time model. Data analytics will allow you to examine the rise and fall of Co2 in your house and determine the high points and low points.

Once you have the high and low points you can then use analytics to determine is there a trend. IE if your high point is always the same time every day what activity is occurring at that time that is changing the levels in your house. Many weather stations now capture the CO2 levels in your home. Some store them for initial analysis but the point of integration would be that system to a broader data analysis system that would allow you to analyze events over time (change over time as well).

In the end I think my proposal is that the IoT is actually a PaaS layer that has three distinct components to it:

  • Bandwidth
  • Integration
  • Data

Bandwidth is both a limiter (more data than available pipe) and an enabler (real time connectivity is available). Integration isn’t always required. Sometimes data can be real time and not require analysis or modification and finally the data itself. Data either needs to be stored and used later, consumed real-time or a combination of both.

Oh IoT what a kettle of fish you’ve opened…

.doc

Scott Andersen

IASA Fellow