Cloud Feedback

Fill it or empty it better with cloud feedback

2010-04-28T22:16:20Z

What do vending machines and garbage cans have in common? They both are in public places; they both need a regular visit, to fill or to empty them; and there are lots of them. There are companies now that will add sensors to them, and use cloud computing services to keep them serviced.

Cantaloupe Systems was founded in 2002 to modernize the vending machine business with communications. They provide vending machine operators a device to install inside a vending machine to communicate every transaction the machine performs. The data is sent over cell phone data networks to Cantaloupe’s data centers, allowing operators to schedule visits to the replenish the machines exactly when needed. The company claims their customers average $35,000 in annual savings per delivery route, reducing their fuel use (and their carbon footprint) by 40 percent by eliminating unnecessary truck travel and removing excess product from the trucks. The data they gather also allows the vending operators to make better choices about products to stock, based on which are selling best in each location. Cantaloupe now monitors 40,000 machines nationwide, with thousands more machines coming online each month. They raised $12m in venture capital this month to finance more expansion.

BigBelly Solar addresses a mirror-image problem: how to efficiently remove what people put into its machines. BigBelly makes a garbage can for city streets that contains a motorized compactor, so that it can receive more trash between visits to empty the can. It powers the compactor from solar cells, so it doesn’t need an electricity connection. It also uses sensors to detect and transmit how frequently it is used, and how full it has become. A cloud computing service records the data and schedules trash collections. The combination of compacting and just-in-time collection saves a lot of fuel and labor; Philadelphia estimates a 70% reduction.

A significant upside to both systems is that breakdowns and problems are detected much quicker, leading to better service. When a vending machine or garbage can detects a malfunction or stops sending data altogether, the operator is notified immediately and a repair can dispatched at once. Taking corrective action even before a complaint is made can save everyone headaches.

What other containers come next? Propane or water tanks for rural homes and businesses? Postal drop boxes? Porta-johns? Feel free to speculate in the comments below.

Cloud feedback serves many masters

2010-03-02T18:42:42Z

Once data about a process is collected and analyzed, the process can be improved. Most of the time, it’s improved in more than one way at once. Cloud feedback applications nearly always have multiple paybacks.

Transportation: safety, maintenance, and fuel efficiency
Buildings: security, comfort, and energy efficiency
Water: water purity, safety from property damage, and water efficiency
Smart grid: reliability, reduced labor, and energy efficiency
Power generation: maintenance, billing, and public relations

The fact that feedback data is stored in the cloud enables additional uses, by allowing multiple organizations to use the data. For example, flood control sensors provide data both to water managers, to tell them when to remove obstacles in the water, and to the public, to provide guidance during a flood emergency. Similarly, the sensors in a smart electric meter enable the property owner to cut energy losses, while saving the utility the cost of manual meter-reading. Most often, cloud feedback systems are designed to serve multiple masters.

One of these ‘masters’ is usually efficiency. When a machine is more closely monitored, there is usually an opportunity to cut its energy use. Sometimes, water or other raw materials can be saved as well. Consequently many cloud feedback applications promote more efficient or sustainable resource use. Efficiency is not a very “sexy” attribute for making sales, however, so other benefits like safety, comfort, or reliability lead the sales pitch.

Home automation for security, efficiency, convenience

2010-02-24T08:18:56Z

There are a lot of home automation systems available today. Some of the oldest go back five or 10 years, and these required a local controller and a special display or a link to the home TV. Also, they were difficult or impossible to control when away from the home. More recent systems are more cloud oriented, employing simpler hardware with browsers and smart phones for control. This article will look at one system that exemplifies cloud feedback, called AlertMe.com.

AlertMe.com began by selling sensors that detect when a door or window has opened or closed, with a communications gateway device that links to AlertMe’s servers on the Internet. The customer can be informed when someone has entered or left the house by SMS message or e-mail or on a website. In addition, they provided special sensors: one is a door bell that can be mounted outside, and another that’s a key fob that you can press to indicate when you’re entering or leaving your house. So their first service was for remote security for a building. From the website you could keep track of comings and goings, set schedules to set or disable alarms, and choose different people or methods to get alarms.

Last year they extended the service by adding what they call a smart plug. A smart plug is a small square device that sits between a wall plug and any electrical appliance. It detects and reports to AlertMe’s servers how much electricity is used by that appliance. It can also obey commands from AlertMe’s servers to turn the appliance power on or off. There’s an additional device that can read the overall home meter. The AlertMe website can show how much power is used at different times both in a browser and on a smartphone, can accumulate that data over months to show trends, and allows setting schedules for turning lights or heaters on or off. Users get information and controls to regulate and reduce their energy use. Some utilities in the United Kingdom, where AlertMe is based, have started to promote the AlertMe solution to their customers to encourage energy efficiency. (AlertMe can also push data into Google PowerMeter — but that’s a topic for a future blog post.)

The AlertMe solution is very low cost compared with other home automation systems. The introductory prices are under 100 pounds sterling. Nearly all of the functions are enabled by software running on the Internet, so the hardware devices are quite cheap, and new features and upgrades are added continuously without the user touching equipment. AlertMe seeks partners like utilities, retailers, Internet providers, and even employers to reach large groups of customers. In a market as crowded as home automation, they retain the flexibility to adapt to opportunities, and tune their offering to partners, by exploiting the cloud feedback architecture.

Cloud feedback provides proof

2010-03-02T18:09:12Z

Using cloud feedback, almost any system can be monitored so that the owner of the system can prove that it performed as expected. This is often a major benefit of cloud feedback. Here are a few examples:

Fat Spaniel Technologies monitors renewable energy sources and records how much power they produce. These records are reported every month or quarter as proof to a government or utility that a certain amount of renewable energy was produced, and therefore that renewable subsidies have been earned.
Green Road monitors the driving habits of bus drivers. The bus company can take these records to its insurance company to show that it has a safety plan in place, and that its drivers are meeting a standard. As a result, the insurance company may offer lower rates.
Demand response devices such as those from Regen Energy also provide useful records. The purpose of a demand response system is to cut demand during the critical hours when a utility has a hard time supplying all of its customers. The detailed minute-by-minute records kept by a service like Regen can prove that demand was cut during the critical times and that a customer merits a payment from the utility.
Water management systems can monitor pollution and verify that discharged water has met water quality standards. The records can also show when and how much pollution occurred, to enable authorities to prosecute those who violate the rules.

There are many other possibilities. For example, a monitor of chemical levels may be required in a drug plant, or even a swimming pool, to prove that safety margins were maintained. A refrigeration monitor could prove that food or wine had been handled properly. Tracking devices on a vehicle can show where it was used, or worn by a convict to show where that person has traveled and whether parole has been broken. We can expect industries and activities that have legal compliance requirements to provide new markets for cloud feedback.

In addition, the need for proof may provide long-term business advantages to cloud feedback companies. As a third party, records from a cloud feedback provider may be considered more trustworthy than records made and held by the customer. Furthermore, once a cloud feedback company has a track record of fulfilling compliance requirements, it will have a significant advantage over new entrants to its market. New customers will value the assurance of past compliance performance, and may be less willing to take a risk on an unproven new cloud feedback service.

Regen Energy saves power

2010-02-18T01:50:31Z

Regen Energy is a Canadian company that uses cloud feedback for energy management. They make a small device about the size of two packs of cards that monitors electricity demand at a single air conditioner or heater, and adjusts the electric use to save energy and money. It publishes the data to Internet web servers to enable additional savings.

Regen calls its device a ‘controller,’ and it can connect to any major source of electric demand. Most often, this is an air conditioner or a heating unit, typically next to or on the roof of a building. The device is connected to the electrical power source so that it can measure the amount of power used. It also sits between the thermostat and the heater or cooler so that the device knows when the thermostat demands more heating or cooling. The device can change the thermostat’s signal, delaying or simulating a demand for heating or cooling.

The Regen devices are designed to communicate wirelessly, both to Regen’s Internet servers and to each other. They coordinate with each other so that they don’t all demand electric power at the same time. (Regen calls this local feedback function “swarm logic“.) In effect, the controllers delay the demands of some buildings until the other buildings can be satisfied. Most utilities levy a “peak power” charge on their corporate customers, based on the highest combined demand of the company. By coordinating power use, the Regen units cut the peak demand and save money on power.

All the data is transmitted via a cell phone data network to Regen servers on the Internet. From there, a website displays it for customers, which offers another way for customers to cut energy costs. They can look at graphs showing how much energy each building uses, and can thereby determine when heating or cooling units need maintenance. They may also discover that some locations are drawing power at times when they shouldn’t (for example a warehouse may be heated all weekend even when no one works there). Regen’s website allows the customer to set schedules and restrict the amount of heating or cooling done during off hours thereby saving more energy. Regen claims typical cuts in energy costs of 25-30%. With entry costs as low as $7,200, even small companies can benefit.

Finally, the company promises to add demand response capabilities in the future. Demand response would allow a customer to keep his heating or cooling units off during periods of high electric demand, in response to instructions from the utility. This can earn significant rebates from the utility.

Regen goes a bit further than simple cloud feedback — its “controllers” are more than sensors. They communicate with each other to coordinate the combined electrical load. They have a clever “proxy thermostat” that effectively and simply switches an air conditioner or heater on or off. Regen combines these functions with an easily accessed and extensible Internet service. The resulting package is simple to install, and can provide energy savings for both small and large companies.

(Note: see comments for updates on controller logic and demand response.)

Watching water from the cloud

2010-02-17T08:03:58Z

Keeping track of reservoirs and rivers is a well-established use of cloud feedback. The use of communications technology to monitor water goes back a long way, and has been transformed by the Internet.

When I lived in Australia in the 80s, I visited a dammed lake in the Tablelands above Cairns. The office of the water authority that managed the lake was about 50 miles away. They kept watch on the water level of the lake with a sensor that floated on the lake and sent a simple radio pulse every few seconds. The time between pulses varied based on the height of the water. The radio signal was picked up in the office by a machine that had a pen and a moving piece of paper. The pen went down every time the signal came in, so the distance between the strokes on the paper conveyed the height of the water. Based on these readings, the water authority would plan releases of water to farmers downstream.

Ten years later I moved to Palo Alto, California. A local creek would flood during especially rainy winters, and the city had set up a number of creek height monitors. It was the mid- 90’s; the internet was spreading and the first browsers were in use, so the creek monitors were displayed in real time on a web page. Residents of the city could see water levels both near their houses and upstream to tell if a flood was coming and they should put out sandbags. The same system works today, with the original display embedded with a webcam on the city’s portal.

Water management is a natural for cloud feedback, since the water is often in locations far from the people who need to manage it. The Australia example was a simple case, with one sensor and one location where people would monitor it. The California case shows a small set of sensors, with results published to affected neighborhoods. There’s a new example from Ireland that takes this to a new level.

Galaway Bay is a large body of water on Ireland’s west coast. In 2008, buoys were placed around the bay carrying sensors for water temperature, movement, and chemical qualities. The data are collected in a cloud computing service run by IBM for the national Marine Institute, which offers public access to the data and to webcams.

The various water authorities and surrounding the bay are able to work from a common base of measurements and data, stored and analyzed in IBM’s cloud. Each authority can customize reports appropriate for their work and publish the results for others to see. (Here’s one on wave height, for boaters and for prospective wave power generators.) One of the big payoffs is that the agencies as well as the private sector can work from a common understanding and common data about the state of the bay.

The vision of the Galway “SmartBay” is ambitious: sensors and computation to address “coastal conditions, pollution levels and marine life,” to “benefit tourism, fishing, aquaculture and the environment,” and to speed reactions to “pollution, flooding, fishing stock levels, green energy generation and the threats from climate change.”

Fat Spaniel for renewable energy

2010-02-06T03:45:04Z

Fat Spaniel Technologies has provided cloud feedback services since 2003. It attaches monitoring and communication devices to wind and solar generators, which send data via cell phone or other networks to their cloud service, which processes and displays the data. They provide a number of ways to use the data. They have grown rapidly, and are expanding to other forms of energy monitoring.

The primary interfaces to the data are:

a Flash-based dashboard with continuously updated graphs and estimates of savings and carbon emissions avoided
a reporting system that can be reconciled against utility billing records or used for renewable energy credits or sustainability audits
an alerting system that can inform customer staff when there are problems with the generator

Their service is simple to understand. When an organization installs a wind turbine or solar installation, it will want to know how the system is operating for as long as it keeps the device. If the device stops producing full power due to a problem (like dirt on a solar cell or a worn bearing on a wind turbine), the owner will want maintenance staff to be informed promptly. If the organization serves the public, it will probably want to let the public know about its use of green energy. The Flash dashboard can be easily embedded on an organization’s web site, and its up-to-the-minute data is an effective attention-grabber. (Public examples include Warner Brothers studio and a California fire station.)

The service is also easy to sell. Renewable energy installers are contractors who generally want to put equipment in and move on. The ongoing service from Fat Spaniel is a natural third-party add-on to their sales. Installing a monitor on each energy device is a simple matter for an installer. For do-it-yourself or small-scale projects, individual devices can be purchased with 5 years of service included. 2500 renewable plants in 20 countries use Fat Spaniel today.

Once set up, the cloud service can be customized, either by Fat Spaniel or the customer, through interfaces in the Flash interface and through a web API for direct data access.

Fat Spaniel has announced partnerships to expand their reach, into energy efficiency and utility-based home services. A new service adds solar facilities management to their products. And I’ve been curious to see how far their solar hot water service will go. This is particularly interesting for utilities that may offer to install, finance, or subsidize solar hot water systems on a large scale. The utilities need to increase the renewable proportion of their energy delivery, and offering a metered end-use service (rather than fuel or power) is a significant departure. The economics of solar hot water are so compelling, especially for commercial or institutional customers, that there may be opportunity for an intermediary, either a utility or “virtual utility”. Metering and monitoring from a cloud feedback service like Fat Spaniel can be an important enabler.

Why use the cloud for feedback?

2010-03-02T18:09:01Z

Cloud computing refers to the placement of computer applications in shared data centers on the Internet. I’ve written and spoken about cloud computing elsewhere, and there are many good summaries online.

Cloud computing provides facilities and systems that benefit many applications — especially feedback applications. Here’s a list of the key benefits:

Internet Communications: The sensors that produce feedback are often far removed from the person who needs it. Normally both have access to the Internet, whether through wired, wireless or cell phone networks. Since cloud data centers have fast and reliable Internet links, they are an excellent point of contact for sensors and users. In addition, the ubiquity of Internet browsers on both personal computers and handheld devices provides a ready-made channel for feedback from the cloud to users anytime and almost anywhere.
Low Ongoing Cost: Today, there is no cheaper place to provide a unit of computing than in a large scale data center using cloud computing technologies. Vast economies at scale have been achieved combining automation, efficient use of power and equipment, and simplified management to minimize operating costs.
Low Start-up Cost: Cloud feedback allows the elimination of on-site controllers and intermediate computers, so that initial costs are much lower. In addition, cloud computing offers compute power in small amounts at very low prices. A prospective user of a cloud feedback solution can begin a small pilot cheaply and grow it economically.
Flexibility and Scale: Cloud computing is designed to handle large competing tasks through a technique called horizontal scaling. This is very well suited to sensor networks which may start out small and grow very large as sensors are added. Each sensor communicates in parallel with the others, in a close architectural match to horizontally-scaled cloud computing, supporting both small- and large-scale systems.
Outsourced Maintenance and Upgrades: The absence of on-site and intermediate controllers means that systems can be more easily upgraded or replaced when obsolete. Like users of other cloud services such as Google Mail or Salesforce.com, cloud feedback customers benefit from upgrades without lifting a finger (and usually without disruptions to service). By comparison with owning and housing computers, upgrading operating systems and software, and replacing equipment at end of life, cloud feedback offers simplicity plus more frequent updates.
Centrally Managed Security: Security for systems which provide feedback can be important, especially if that feedback is the basis for automatic processes. For example, if feedback is used to automatically set thermostats or control machines, it’s important that the feedback loop not be compromised by intruders or software bugs. At first cloud computing might seem more vulnerable since cloud facilities are available over the Internet. However the centralization of facilities in the cloud permits a single group of security experts to maintain the security of the software and communications, with the cost spread over large numbers of users. When each user is responsible for the security of their own on-site equipment, the security levels are commonly lower than centrally-managed systems.

Thanks to cloud computing, system costs are much lower than even a few years ago. Meanwhile, the number of potential feedback applications has grown due to advances in sensor technology and the increased value of energy and water efficiency. This blog will highlight many examples of cloud feedback, with particular attention to the ways they exploit the cloud.

GreenRoad cuts accidents and saves fuel

2010-01-29T08:19:55Z

GreenRoad is an interesting example of cloud feedback. GreenRoad installs a device inside a car, truck or bus to monitor the driving and fuel consumption of the vehicle. Plugging into the vehicle control systems that are present in most modern vehicles, it records speed, driver actions, and fuel consumption. It analyzes the steering, braking and acceleration of the driver, and it notices frequent turns or lane changes or sudden acceleration or braking. The device matches the data against a claimed 120 patterns to evaluate fuel efficiency and driver safety. It displays these quick calculations as simple green/yellow/red lights for the driver.

It also conveys all this information via cell phone data networks to servers in Greenroad’s data center. Drivers get periodic reports and personalized driving tips. Using GreenRoad’s portal, the owner of the vehicle (e.g. a bus fleet manager) can see trends in the data as it accumulates. The information is graphed to show which drivers are the most efficient or the most safe, and the ones that are getting better. The combination of local feedback to the driver, and cloud feedback to both driver and owner, results in significant improvements in energy efficiency and safety.

Like other cloud services, GreenRoad has low up-front costs and a reasonable monthly charge (currently about $30 per-driver per-month). They claim that savings are realized from the first month of operation in most fleets, both from reduced crash costs and lower fuel bills. Their overview presentation cites truck fleet savings of $1500-$4000 per truck per year. Follow-on effects are also significant. Insurance companies like AAA and Hartford have partnered with GreenRoad to promote the service to increase safety, which may result in lower insurance rates. GreenRoad also supplies measurements of carbon emission reductions from fuel savings, which its customers frequently advertise.

GreenRoad is in use or being trialed on several truck and bus fleets in the US and Europe, including city bus fleets, the Ryder truck fleet, the UK Ministry of Defense, and some US school bus fleets. It’s even being trialed with teenage drivers, including insurance rebates for participants.

Side note: Samuel Chiu and Eric Carr are conducting a research project at Stanford that hopes to achieve similar functions with an iPhone app. They use the location and motion information from the iPhone to infer the actions of the driver, rather than tapping into the engine directly as Green Road does. No results are published yet.

What is cloud feedback?

2010-03-02T18:08:43Z

Whenever we use a machine, we measure something to get feedback about how it is performing. When I drive my car the speedometer gives me instant feedback on how fast I am going. I get slower feedback when I refill the car and find out how much gas I’ve used. If I run a store, my cash register will tell me what is selling and what I need to restock. When I use an exercise machine, it may measure how long I’ve exercised, and when I step on scales I get a measurement of how effective the exercise has been been.

The more things I measure and the more frequently I measure them the better the feedback; and the better the feedback, the better the results of whatever it is I’m trying to do. Naturally it makes sense to put a computer in the loop. A computer can keep track of the feedback automatically, it can compare feedback with the past, and it can even tell other machines (“actuators”) to take action based on the feedback.

Conventional feedback diagram

As recently as five or ten years ago, it was very costly to computerize feedback systems. It involved placing a computer at the same location as the thing being measured, usually with a wire installed to connect to the sensor, and with a custom-designed user interface to deliver the feedback to a human being. If off-site access to the feedback data was needed, an additional remote access network had to be employed. Now things are different. Internet access is ubiquitous, thanks to wired, wireless and cell phone networks, so the computer can be located almost anywhere. Today, the feedback interface is easy to to build, since any Internet-capable device with a browser will do.

When a computer is in the feedback loop, and it runs in a data center located on the Internet, I call that “cloud feedback”. The computer is “in the cloud” and its program will deliver feedback about any machine or process in the real world. Cloud feedback computer applications are growing rapidly, primarily due to reduced costs. The cost of computing in the cloud has dropped dramatically (between 10X and 100X) in the last ten years. Meanwhile sensors that can detect things in the world have also gotten much cheaper and smaller.

Cloud feedback diagram

The purpose of this blog will be to identify a number of examples of cloud feedback systems, and to look for common patterns in their purposes, design, and implementation.