Wireless Sensor Contextual Computing

Wireless sensor packs reduce the compute complexity of a local IoT network by processing the information closer to where it is most useful.

No matter which research firm you look to, you’re going to find the same answer—the market for the Internet of Things (IoT) is going to be huge. Gartner, for example, forecasts a 30-fold increase in Internet-connected physical devices by 2020, reaching 26 billion installed units. IDC, on the other hand, believes the installed base of things connected will be 212 billion by the end of 2020. The much anticipated “billions of things” connected to the Internet will provide numerous benefits to consumers and businesses alike, but can also become unmanageable very quickly based on the sheer volume of information that needs to be processed. The majority of IoT nodes created will consist of wirelessly connected sensors that sit at the edge of the Internet. These devices monitor activities or environmental conditions and report basic data or status for a local environment. This information, captured by wireless sensor packs (mix of specific sensors), can be processed at the sensor node and broadcast commands, as opposed to raw data, to a mesh network using the 802.15.4 standard. Specific controllable devices in this local mesh will recognize the types of commands issued from a sensor pack device and execute the command. This contextual processing provides a power-efficient failsafe backup system to ensure local control isn’t lost when the broadband Internet or Wi-Fi connection goes down. This article will provide readers with a technical overview of how this works and the ways in which the methodology reduces the compute complexity of a local IoT network by processing the information closer to where it is the most useful and in the process, generate a lot less Internet traffic and IoT devices.

Figure 1: Information connectivity driving the number of nodes in the IoT

Close Up: 802.15.4 Standard Technology Running ZigBee Protocols

With a multitude of different devices connecting to the Internet to provide information connectivity, a local network will need to be formed to manage sensor and control nodes for a given context. Utilizing 802.15.4 technology running ZigBee protocols via a mesh network will be one popular method to connect devices. Within a ZigBee mesh network, there are edge routers (gateway to the cloud), routing nodes that form the mesh for multi-hop connectivity that could also act as control nodes and host nodes at the network edge that contain sensors and potential actuators.

Figure 2: Typical IoT nodes within a building or home, allowing access to the cloud

Typically these sensor nodes send raw information such as motion detection within a timeframe, ambient light, temperature or other information to a sensor aggregator that will then connect to the gateway or even directly to the next-generation smart gateway. The smart gateway usually has the network intelligence to decide what to do with this information. Sending it out to the cloud to a service provider to manage things like energy usage, security or remote monitoring with the ability to control all that from a smartphone is an option. With smart gateway Wi-Fi connectivity directly to the smartphone, you may be able to take action locally based on the information received. With more recent gateways from a service provider, local decisions can be made to take action based on specific programmed conditions. For example, it’s possible to have the lights turned off at a certain time when it is determined that no one has been around for a few hours. There are many graphical user interface (GUI) interations required to set these types of predetermined conditions in conventional systems.

The management of IoT devices can be used to save energy; however, it takes many routing nodes and typically a gateway to create controllable definitive actions. A relatively complex program has to be created to be run on the gateways to process sensor information and then take appropriate actions to control local IoT devices in an efficient manner. The same applies for cloud-based services. The farther the physical distance from the contextual event to analyze sensor information, the more complex it is to create an efficient control system. In the case of cloud-based IoT services, the connection to the IoT device will be broken if the service provider modem has lost broadband connection or the local gateway loses Wi-Fi connectivity. Essentially all local control can be lost as well as an opportunity to save energy. A third level of control can be made possible with the advent of smart sensors using a contextual computing methodology.

Close Up: Wireless LED Lighting

One of the easiest and most accessible ways to install IoT devices in the near future may be the approach taken by wireless LED lighting. Similar to the first method, this approach also has its drawbacks and advantages. Starting with the positive, this technique presents a great opportunity to do things like dimming the lights and turning them on or off through the convenience of a smartphone. With that, though, there need to be specific actions to save potentially wasted energy. The LED bulbs could be programmed to act according to a schedule running on a gateway or cloud services. A smart gateway has the ability to do this automatically, whereas cloud services could manage the actions assuming all wireless connections to the cloud are useable. A more interesting approach would take advantage of passive information from a contextual setting to control devices without the need for scheduling or cloud service intervention.

With ZigBee Light Link-enabled wireless LED bulbs, a wireless mesh network with a lightweight protocol stack, is formed by screwing in a handful of LED light bulbs in a room, house or office. No IT technician is required, as the mesh network is self-created and self-maintained. In fact, more devices can be added at a later time automatically. Smart wireless sensors can connect into this mesh network to provide information to the smart gateway or out to the cloud services management to take appropriate actions.

Figure 3: Mesh Network formed by wireless LED lighting

Smart sensor nodes typically have an MCU with onboard flash and serial connections to connect a group of relevant sensors and the required connection to a wireless module. For home lighting control for instance, a group of sensors in a home environment may include ambient light and motion detection. Once the sensor node has registered on the smart gateway network, any information gathered will have contextual information for a given location. The gateway or cloud services could assign some local controllable devices such as LED bulbs to use this contextual information with some level of priority. Management of these specific LED bulb devices still resides in the gateway, cloud services or a smartphone.

With the introduction of low-cost, low-power MCUs containing flexible onboard flash sizes, some of the analysis of the sensor information can be done efficiently and directly at the sensor node. The information gathered in these smart sensor nodes can be processed locally to provide contextual relevance. Instead of sending raw information to be processed remotely, the smart sensor could form a series of commands based on the local sensor analysis. This smart contextual sensor information can take action directly to local devices assigned to it without the need for an intervention from a gateway or cloud service.

How Does this Work?

For example, if there was no detected motion for the last 30 minutes in a given area and the sensor could read some level of light, the sensor could directly send a command to dim the local LED bulb assigned to the contextual area. The smart sensor can check at a later time to see if any additional motion was detected. If none is detected in a given timeframe, a command could be issued directly to the LED bulb to turn completely off. These smart sensor commands will become valid to the LED bulb only if the gateway has not detected any recent commands from other sources with overriding priority, enabling the LED bulb to act autonomously to a smart sensor command point. The action of the LED bulb to react to the direct command from the smart sensor could also transmit this information to the smart gateway or out to the cloud. If a cloud service has initiated a service routine such as “turn on lights at a low level between a fixed timeframe for security concerns,” the sensor’s autonomous control will be overridden at the LED bulb. A cloud services command could be sent to the LED bulb to ignore any direct sensor control information due to a higher level of overriding priority. Once the cloud service action has been completed, the priority override of the smart sensor to the LED bulb would be canceled.

Close Up: Energy Harvesting

One form of the smart wireless mesh sensor could be realized using energy-harvesting techniques, creating a low-cost smart sensor node without the need for wired connectivity or replacement batteries. These devices could be applied physically anywhere for contextual computing.

Figure 4: Energy Harvesting Sensor connectivity to the Cloud

Precisely placed wireless smart sensors can take advantage of location-based priority information such as installation by a doorway to monitor traffic within a floor of a building or a specific office area. Wireless motion sensor information could be transferred to the mesh network running the ZigBee Light Link protocol stack with the help of an energy-harvesting PMIC with a solar cell. A piezo-based energy generator converts physical motion (such as walking on a carpet) to electrical energy to power the wireless smart sensor. Floor-based sensors built within a piece of carpet could use the piezo-based energy harvesting method to monitor traffic and only send information when someone walks on the high-traffic area.

Key considerations for a useable wireless sensor system include:

  • A low-power MCU with appropriate onboard flash memory requirements that also incorporates quick start-up capabilities and sleep-mode power savings with a real-time clock (RTC)
  • A PMIC with a solar cell energy input or piezo energy input to store charge in a super cap or recharge a battery for long life with low maintenance
  • A low-power 802.15.4 transceiver

In an energy-harvesting environment, joules are now monitored instead of voltage, current or power, within a smart sensor network. A bucket of energy required to perform a task is the latest paradigm shift.

Figure 5: Energy Harvesting Sensors system connectivity

With the flexibility to place these smart contextual computing sensors exactly where needed, a more efficient system can be created to passively manage IoT-controllable devices such as LED bulbs without the need for complex analysis programs running on a smart gateway or cloud-based services. No wires. No battery changes. No maintenance. This equals lots of simple to manage energy savings.

As the IoT continues to gain momentum, the question is no longer focused on how many billions of connected devices are in our future. Instead, the focus now remains on how to manage all these connections efficiently as this technology comes to fruition.

Spansion_Wenzel_Pic_onlineJames Wenzel is the director of technical business development and strategy for the Industrial and Consumer Micro-Controller group at Spansion. James graduated from the founding class of micro-electronics bachelor’s degree program at Rochester institute of Technology. He has written numerous papers with multiple industry presentations for semiconductor design, packaging and manufacturing. James currently has 7 patents granted with 4 pending.

Spansion Handa_Formal_OnlineDhiraj Handa is the senior vice president and general manager for the Mass Market, Industrial and Consumer Marketing business unit at Spansion. He has been an active contributor to the semiconductor industry for over 25 years, with experience in semiconductor design, operations, product roadmaps, marketing and managing global corporate functions. Dhiraj began his career as a design engineer at IBM and has an MBA from St. Edwards University and a BS in electrical engineering from the New Jersey Institute of Technology.

Share and Enjoy:
  • Digg
  • Sphinn
  • del.icio.us
  • Facebook
  • Mixx
  • Google
  • TwitThis

Tags: , , ,