Here Comes the Swarm…




Not to mix metaphors, but when the “swarm” is a distributed network of sensors checking multiple personal health and fitness indicators, we could be opening a can of worms.

“Here Comes the Swarm” isn’t a B movie, it’s a prediction. Yes, the swarm really is coming. But this swarm won’t be teeming with killer bees or ravenous locusts or harbingers of the zombie apocalypse. Instead it’s made up of a multitude of sensors—covering our bodies and monitoring a wide range of our daily activities and various bodily systems to promote our wellbeing. And the fact that we are rapidly moving from today’s reality of a handful of sensors integrated into a single platform such as a smartphone, to tomorrow’s vision of dozens or perhaps even hundreds of sensors placed across a person from head to toe, engenders a host of implications that we as an engineering community are only starting to contemplate.

sensors_image
Figure 1: Which device will have the authority to evaluate and prioritize data from multiple sensors?

Head and Shoulders (and Knees and Feet and…)

Unfortunately, there is no single ideal place for wellbeing sensors. Some sensors need access to the skin to measure things like heart rate, while others need access to the air, to measure things such as air quality. Some work better on the feet and others on the waist, chest, or head. Therefore we can assume that, as sensor costs come down, we will start to see sensors distributed across our bodies—each placed in their respective near-optimal locations. Some examples already exist, for instance the Garmin Foot Pods and the Mio Velo and Link heart rate monitors.

As sensors become more widely distributed, it will ultimately make sense for them to be wirelessly networked together, probably using a communications protocol such as the low-energy Bluetooth Smart standard. Once that happens, we will find cases of duplicated capability, which will create the opportunity to select among sensors based on their location and resulting quality of data. For example, a wrist is a difficult place to get accurate step counts because arms move even when people are not walking. A smartphone in a pocket does not have that problem, but relying on the smartphone might risk missing data when the user is walking and the phone is not on the user, perhaps being charged.

Accuracy and Power Considerations

If we think of the smartphone and a wrist wearable as part of a personal sensor swarm, then swarm fusion can determine which device, wrist or smartphone, is giving the more accurate data and choose accordingly. Similarly the ear is a much better place to monitor heartrate, so when a person is using HRM enabled earbuds, the swarm can use data from them rather than from a wrist mounted device.

Another consideration is power. With an intelligent swarm of sensors, the swarm can decide which sensors are not required at any given time and power them down. Again, for example if a person is walking with both a wrist device and smartphone, since the smartphone has a larger battery and superior quality data, the swarm can shut down the pedometer function in the wrist device and rely on the phone, at least until the user removes it from their pocket.

Having a distributed network of sensors creates other issues. For example, which device (or set of devices) makes the decisions about data quality or power or whether or not a particular sensor is functioning properly? A likely starting point would be for one device, perhaps the user’s smartphone, to be the “main” device and to have the authority to evaluate and prioritize data from multiple sensors. These might include sensors integrated into the smartphone itself, as well as other sensors distributed across the body. For example, in the simplest case a phone might look for a foot pod, and if it finds one in the swarm network it could turn off its own pedometer to save power.

Decision Distribution

Eventually the decisions will be distributed, at least partially. Devices will broadcast their capability, and, based on the capability of those around them, will either shut down or continue sensing and transmitting data. Extending our pedometer example, a foot pod might broadcast that it is currently counting steps (with high accuracy) in which case both the user’s smartwatch and smartphone would halt their step counting function. If the foot pod stopped broadcasting, then both the watch and phone would start. Then, based on that scenario, the watch (knowing that its step counting function is less accurate than the phone’s) might halt its pedometer.

Accelerating the Process

Managing either centralized or distributed decision-making and control will initially be challenging for the industry. If the development of networked personal sensors rolls out in a similar way to past transitions of this nature, initially the results will be painful and frustrating. The first stage will be closed systems from specific manufacturers in which all of their devices communicate with each other but not with devices from other manufacturers. That approach will help manufacturer margins and improve the user experience within the closed system, but will also engender user frustration over their inability to include devices outside of the system. The second stage will be an open network, which tends to commoditize the hardware, but will ultimately work best for users who now have access to everything and can choose best-in-class or best-value, depending on their needs.

One hopeful trend that could really accelerate this process is the role ARM is playing. We all know of ARM as the company that develops the processors that power the mobile world—from the high performance processors in smartphones to the low power M0+ that powers Bluetooth beacons. But ARM is also pulling together the companion IP, software and eco-system that the distributed world of IoT and wearable swarms will need.

Prepare for the Swarm

Elements of the swarm are already starting to appear (Figure 1). There are heart rate chest straps, optical heart rate wrist bands, and foot pods that feed to a smartphone. There are also speed and cadence sensors on bicycles and we can expect to see more sensors showing up on various forms of exercise and sporting equipment. The devices in these systems don’t yet have a network actively choosing data from different sources, or managing power consumption across the system, but it seems likely that soon they will. More and more devices will be added, with increasing levels of wireless connectivity and the resulting opportunities to manage device functionality, data quality and battery life.
So we all should prepare for the swarm. It will be here sooner than we think.


exec-tim-saxeTimothy Saxe (Ph.D) joined QuickLogic in May 2001 and has served as Sr.Vice President and Chief Technology Officer since November 2008. Prior to this role, Dr. Saxe served as the company’s Chief Technology Officer and Sr. Vice President, Engineering from August 2006 to November 2008 and as Vice President, Software Engineering from May 2001 to August 2006. From November 2000 to February 2001, Dr. Saxe was Vice President of FLASH Engineering at Actel Corporation, a semiconductor manufacturing company. He holds a B.S.E.E. degree from North Carolina State University, and an M.S.E.E. degree and a Ph.D. in electrical engineering from Stanford University.

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