At last week's Internet of Things Developers Conference held in San Jose demonstrating the value of IoT was clearly top-of-mind. There were predictions of the number of devices that would be part of the IoT ranging from 20 billion to 50 billion. One speaker mentioned that a 1% reduction in maintenance may increase profits by up to 15%. Another presentation described a smoke detector that not only detects the smoke and calls the fire department but also knows who is in the house and what room they are in rather than just beeping, it can provide verbal instructions like "There is a fire in the kitchen, so leave immediately and go out through the garage door. The fire department has been notified and is on the way." And for young kids, it uses Mom's voice to talk to them so they don't get scared. Obviously a highly controlled interconnected multi-device solution, but it can surly be done today!
A clear fundamental theme of the conference was security. Most every presenter mentioned it. Notably a panel on IoT Security offered: "We cannot make a device that's fully secure, because there are always new and different security attacks. When we see a new attack we must do something different to cover that security hole, so being able to update the device is an ABSOULTE MUST."
Other – There were several miscellaneous items I found interesting.
- There were multiple Custom SOC vendors.
- One custom CPU vendor had a low-end 32-Bit CPU that was built on only 7700 gates.
- A new compiler is available that can easily enable custom instructions to be used by defining those custom instructions to the compiler. During code generation they will be used correctly, meaning you don't have to change your C/C++ code to use fully custom instructions.
- Dream Factory has a fully Open Source Server software that enables RESTful API's to connect to IoT devices. This server software is a LAMP solution that enables multiple database models.
- Qualcomm promoted the idea that all the technology found in smart phones is needed for IoTlike the multi-sensor aggregation, security, audio and video.
One of the areas that I found especially fascinating is the state of computer vision and the fact that inexpensive IoT devices can use it today. Visual data offers perhaps the richest source of information about the real world: people, places, and things. The low cost camera sensor and DSP technology makes computer vision cost effective. Computer Vision can enable a device to know your mood, know who is nearby, enable a computer to map out an area and plot a course to some new destination. One of the products that is using it is a new automatic vacuum cleaner that maps out the room and only vacuums any given spot once, not like the "old school" robotic vacuums that wander around randomly and just about wear out your carpet to get it clean. Having the DSP in the IoT device means that vision data need not be sent to the cloud to be processed, but only the metadata found the within the vision data.
The perspective of Cloud provides the storage and analytics was bantered around a bit. Yet one application that was referenced is that an Oil Rig generates about 2 terabytes of data per day and 99% of it is discarded as useless. Datalight has seen similar use cases in our work with airplane engine data recorders which generate about 1 terabyte per hour. The important data to communicate in the moment is what's changed, but the details can become important when disaster strikes and you need to find the root cause. In those use cases, both local storage and remote (or Cloud) storage can work together.
Building an IoT device on which data is critical? Reliance Edge may be just what you need.