When devices can sense and communicate via the Internet, they can go beyond local embedded processing to access and take advantage of remote super-computing nodes. This allows a device to run more sophisticated analyses, make complex decisions and respond to local needs quickly, often with no human intervention required.
Let’s take a look at the most common use cases for the Internet of Things.
Pervasive Remote Tracking/Monitoring and (if needed) Control & Routing (TCC&R): This refers to remote tracking/monitoring and, if needed, command, control and routing functions for tasks and processes today usually done manually, or, if done remotely, that require additional infrastructure. For example, in most homes today, it’s a manual process to turn on and off certain lights, set temperature zones and turn on and off a washing machine. In the future, doors, windows, electrical outlets, appliances and many other types of standalone equipment will become “smart” with a unique ID. Those smart devices can then be connected via wired or wireless communication, allowing a user to monitor his or her house remotely, change settings on a refrigerator or washing machine and control household tasks through a laptop or mobile phone. In fact, there are some services offered today by security or Internet service providers to do exactly that, but on a much smaller scale and with fewer capabilities than we’ll see in the future.
Asset Tracking: An extension of these kinds of services is asset tracking, which today is done via barcode and a variety of manual steps, but in the future will leverage smart tags, near-field communication (NFC) and RFID to globally track all kinds of objects, interactively. The word geo-tagged is now being used by some companies to refer to this class of applications. In a future scenario, a user would be able to use Google Earth to track anything with an RFID tag. Alternatively, your refrigerator could keep track of your smart-tagged groceries and tell your cell phone app you’re low on a certain item. If your bag of frozen vegetables can have a smart tag, other objects such as valuable cars, jewelry and handbags could too, and they could be tracked via the Internet and also take advantage of a variety of available web-based applications. Some telehealth-related services also belong in this category. The graphic shown below gives an example of how even a pill’s progress through the human body could be remotely monitored.
Process Control and Optimization:
This is when various classes of sensors (with or without actuation capabilities) are used for monitoring and to provide data so a process can be controlled remotely. This could be as simple as the use of cameras to position boxes of various sizes on a conveyer belt so a label machine can properly apply labels to them. This task can be done in real time by sending the data to a remote computer, analyzing it and bringing a “command” back to the line so various “control” actions can be taken to improve the process … without any human intervention.
Resource Allocation and Optimization:
The smart energy market provides an ideal example of this use case. The term smart energy
has been used in many ways, but it basically refers to accessing information about energy consumption and reacting to the information to optimize the allocation of resources (energy use). In the case of a household, for example, once the residents know they’ve been using their washing machine during peak hours when the grid is most constrained and the cost of electricity is at premium, they could adjust their behavior and wash their laundry during non-peak hours, saving money and helping the utility company cope with peak demand.
Context-aware Automation and Decision Optimization:
This category is the most fascinating, as it refers to monitoring unknown factors (environmental, interaction between machines and infrastructures, etc.) and having machines make decisions that are as “human-like” as possible … only better!
Here’s a personal example from Kaivan’s past that can help illustrate this: “When I was a young engineer, I worked on a traffic collision avoidance system (TCAS). In that system, when two airplanes were approaching each other on a collision path, the ‘machines’ in the two airplanes would take over. The system first would send an audible warning to the pilots about the danger ahead, while at the same time communicating between the two planes and deciding how each plane should move to avoid a collision. The assumption was that if the two pilots were warned and were in control to make quick decisions, they could both decide to make turns that would still cause a crash.”
There are a whole host of new technologies available today and in development that could allow vehicles to communicate with each other as well as with a central control unit. These smart vehicles also could sense the road, traffic signs and lane markers and, using GPS and a communication link, avoid incoming traffic, avoid accidents around a curve or, in conjunction with the central control unit, avoid going over a distressed bridge on the verge of collapse.
Remote patient monitoring is another example relevant to this use case. For instance, imagine an implantable sensing node that tracks biometrics and sends a signal regarding an abnormal readout for an elderly patient. If the patient doesn’t respond by taking a medication, the node could place an emergency call to a contact from a list, and if there’s no answer, call a second contact, and finally, if no answer, contact a monitoring clinic or quickly provide other emergency assistance. Another example is continuous monitoring of chronic diseases to help doctors determine best treatments, with minimal human intervention.
Requirements common to all of the use cases above include:
- Sensing and data collection capability (sensing nodes)
- Layers of local embedded processing capability (local embedded processing nodes)
- Wired and/or wireless communication capability (connectivity node)
- Software to automate tasks, and enable new class of services
- Remote network/cloud-based embedded processing capability (remote embedded processing node )
- Full security across the signal path
In the factory automation example mentioned above (applying labels to boxes), a camera detects information using a charge-coupled device (CCD) sensor (sensing node), the collected data is then communicated to an embedded processor/controller (embedded processing node) using wired or wireless communication technology (connectivity node), a decision is made by the remote server (remote embedded processing node) and communicated (connectivity node), which causes a mechanical action to take place that corrects the situation.
For a Context-aware Automation and Decision Optimization example, we can use the example of a smart car, using its active safety radar system (sensing node), in conjunction with image processing cameras (sensing nodes), it communicates with an embedded processor (embedded processing node) in the center stack of the car to make an appropriate decision regarding danger ahead. Or, the vehicle could leverage its built-in GPS and wide-area-network (WAN) wireless communication capability (connectivity node) to pass along information to a central processing server on the network (in the cloud) (remote embedded processing node) that could then make the car aware of the information it had just received, from the sensors of an unstable bridge (sensing node) on the road ahead, that was being pounded by rain flood and loosing its structural integrity, and hence guide the car to a different route to avoid danger.