Website Logo. Upload to /source/logo.png ; disable in /source/_includes/logo.html

OpenSensors.IO

News and Updates about OpenSensors.IO

Transcript of Evidence Based Design Webinar

Edited transcript of the OpenSensors Webinar on “Evidence Based Design of Workspaces” Mar-28-2017

Our panelist are: * Arjun Kaicker of Zaha Hadid Architects. An architect and workplace consultant with more than 2 decades of experience in workplace design, he brings a real passion to his work and wants to create workspaces that directly respond to the needs and aspirations of his clients. * Yodit Stanton, CEO of OpenSensors. She has spent the last 15 years as a data engineer building large scale data processing and machine learning technologies in financial trading systems. She has been working with IoT data for the last 3 years. * Sean Murphy, CEO of SKMurphy, Inc. is an advisor to OpenSensors. Acting here as the “voice of the audience:” he poses questions to the panel as webinar attendees type them in.

Workspace Design Decisions Can Now Be Informed by Evidence from Sensors

Yodit Stanton: Thank you everyone. Welcome. This webinar covers data driven design or evidence based design. Essentially what that means is using data from sensors and other things around enabling people to understand how space is being used and also the design of the space. A lot of the trends we’re seeing, is in especially in building occupiers, remotely monitoring the buildings. There are also starting to use these data sets to inform the design and inform the future planning.

Arjun Kaicker: At Zaha Hadid Architects we’re really interested in the potential of sensors. Architects and designers have always struggled to really understand client needs in office projects. An office design is the opposite of a residential design. In a house there is a family with a few people who are the primary users of the building. In a office there are hundreds, possibly thousands, of users with very diverse behaviors and contradictory requirements.

Sensors provide a really powerful tool, for understanding workplace needs as we never have really been able to before. We no longer need to rely on assumptions and preconceptions of how people work, or on benchmarking, or—even worse—copying what other successful companies do with bean bags and foosball. Sensors really give us the opportunity to take the guess work out, take the assumptions out and to understand the real needs of workplace users.

Yodit Stanton: At OpenSensors we see three drivers for work space design:

  • The recognition that desks are underutilized in many offices is driving new forms of desk assignment where real estate costs are high. In London for example, the average cost of a desk is somewhere between 13K to 15K a year. It’s leading managers to ask, “How many desks do I really need? If I have 100 employees in my group do I really need 100 desks or will 80 be sufficient?”

  • Matching the right mix of meeting room—and break room—configuration options with employee needs for collaboration and communication: how many phone booths do I need? How many small rooms and how many large rooms. Are break rooms being utilized. Meeting rooms are often a point of contention and while booking systems have helped there are still challenges with someone reserving a block of time and then not using it.

  • The environmental aspects of the workplace, in particular noise and air quality. Both impact employee wellness and productivity. I think the Well Building standard (see https://www.wellcertified.com/system/files/WELL%20Building%20Standard_v1%20with%20January%202017%20addenda%20.pdf ) is one example of this. Another is manager’s concerned about noise levels in desk areas, what’s appropriate for the tasks they are performing.

Arjun Kaicker: Great points Yodit. Workplaces are expensive so we don’t want to waste space. We don’t want meeting rooms that are being underutilized or amenity spaces that are not as popular as people predicted. But the flip side is that we often find some spaces over utilized in the workplace so that they are not available when needed. In that case the problem isn’t about waste, it’s about people not being able to do their job properly. When people can’t communicate and collaborate properly because they can’t find a meeting room at the right time, it can have a real effect on business efficiency and productivity.

Sean Murphy: Arjun, I have a question for you. Yodit presented figures that it was 13 to 15 thousand pounds a year per desk in downtown London. I would think that the energy cost at most a 10% of that—perhaps even less than 1%—yet we’ve spent more time trying to instrument the energy usage than we have the space usage. Why do you think the space usage monitoring has lagged?

Arjun Kaicker: That’s a fantastic point. With energy a few sensors can capture the total use, while you may need more to get a finer grained understanding. I think space usage monitoring has lagged because you need many more sensors to be able to monitor it as carefully. Building management systems have enabled both the monitoring and adjustment of energy usage. Traditional reservation systems have not incorporated data from meeting room occupancy sensors so they have only been able to allocate but not measure actual usage. Swipe card systems that monitor every entrance and exit can be used to assess total occupancy but give very little detail on usage. I think that occupancy monitoring and utilization assessment are going to catch up with energy monitoring.

Real Goal Is to Enable Employees to be More Productive

Arjun Kaicker: We cannot lose sight of the fact that it’s the people and not the real estate or the energy that are asset, and the most expensive cost. Depending upon their skills and experience the total cost of the people is probably seven to ten times that of the space they occupy and energy they consume in an office setting. If a better designed workplace can increase productivity by five or six percent that’s equivalent to half of your real estate cost And, if you can, from getting a better designed place, from getting a better designed workspace. If you can increase productivity, the efficiency by 12 to 15% you have paid for your whole building—that’s where the real savings can come.

Space utilization studies

Sean Murphy: Currently how are folks monitoring the workplace? And how can using sensors make it more effective?

Arjun Kaicker: Today you can look at the swipe card data to see how many people are occupying a building but that doesn’t really tell you anything once they’ve gone into the building.

You can interview people or hand out surveys to ask them what they see working well and not so well. But this can end up being very subjective. As a workplace consultant, the only technique that I’ve used that did more than scratch the surface is a space utilization study.

For a space utilization study we start at one end of the building and walk to the other end, floor by floor. We walk to as many desks as we can within an hour and then turn around and start again. For eight hours we mark down what’s happening at the desk or what’s happening in the meeting room.

With one person in a week you can get good coverage on about a hundred desks: you have a snap shot for the week of how those desks were used. This is a very very useful complement to interviews and satisfaction surveys for an organization. The problem is that it’s just one week and it would only be for that second you walked past the desk during the hour. So, if someone happened to get up to use the restroom for five minutes while you were walking by you would mark the desk unoccupied for an hour. It’s a time consuming expensive method: to cover 1,000 desks you would need a least five people full time for a week.

Sensors Are Replacing Manual Site Surveys

Arjun Kaicker: Sensors have massively cut the cost of carrying out this kind of data gathering and the amount of time it takes. We have also found that although space utilization studies were very useful to give a kind of general picture, they weren’t that persuasive because people knew were only done over a week or they knew that actually, it was only a snapshot within a second within that hour. And, I think that sometimes managers and executives were a bit suspicious of them: it’s much easier to buy into requirements based on the very rigorous data from sensors.

Sensors Are Not a Substitute for Conversation

Sean Murphy: How do you tell if people have tried but failed to find the certain type of space? I needed a phone booth but I couldn’t find it? I needed a five person conference room and ended up in one that holds 20.

Arjun Kaicker: You cannot tell directly but you can be pretty sure it’s happening if we see 100% utilization of a particular resource. Normally 80% to 90% is about the limit before problems start to emerge.

Yodit Stanton: Sensors are complementary to satisfaction surveys and in-depth conversations. I would never say, even as a sensor company, that they can replace the in-depth conversations that you need to have with employees.

Arjun Kaicker: To give an example of that, a few years ago I had in client on the West coast of Canada. We did a space utilization survey—sensors were not available then—and we did a questionnaire. In the questionnaire people kept on saying ‘I can never find a meeting room when I need it’ but the space utilization study found that meeting rooms were used 30% of the time. We drilled in to see if there were any particular meetings rooms which had high use—maybe these were the ones people were complaining about—but answer was no, the highest meeting room usage was 60% of the time.

We dug in some more and did some in depth interviews and discovered that they were doing a lot of audio and video conferences with colleagues on the East Coast of Canada. There’s a four hour time difference and now it became obvious. For half the day in the window where their workday overlapped the over coast, these meeting rooms were booked solid. We were able to start designing to meet this need once were were able to reconcile the utilization data with the survey data and insights gleaned from in depth conversation.

How Are Sensors Changing the Architect’s Role?

Yodit Stanton: How do you see the role of architects changing? My understanding is that architects used to design the space and deliver the project and then move on to their next project. Now with these sensor networks actively collecting occupancy data and other systems generating live data, how do you see the role of architects changing with this ongoing stream of information about how the client is taking advantage of the design?

Arjun Kaicker: Sensors can make a really big difference to the way that architects design space because now we’re designing more for flexibility, for adaptability in the future. We’re not just looking at designing in day one. Sensors provide data that allows us to move beyond rules of thumb and best practice, to they enable us to understand client needs in much more depth so that can we design something that is better for them. Sensors are holding us more accountable to clients for the impact and usefulness of our designs

How OpenSensors Helps Evolution of Smarter Buildings

Yodit Stanton: We are a technology company, most of our customers come to us with a project and ask for help selecting the right sensors, managing their installation and integrating the data streams they emit with existing tools and information systems.

We really do three things:

  • Find and evaluate sensors for inclusion in sensor networks. We are always looking for new options that provide better battery life, better range, higher reliability, new protocols, or otherwise extend the set of capabilities of what we can offer. This often involves establishing working relationships that allow us to be knowledgeable but vendor agnostic.

  • Build and manage sensor networks that collect data and turn it into useful information. We spend a lot of time establishing partnerships with other firms that allow us to make proposals that include the installation and ongoing maintenance of not only software and cloud stack but on-site the hardware—primarily sensors and gateways.

  • Integrate with existing systems to provide them with useful information: this is what people really care about. Can you extend the capabilities of tools and systems I am already using to take advantage of this sensor network that has been installed. We believe that the trend toward smarter buildings will continue to be evolutionary, so we strive to interoperate with and extend the uses for tools that our clients are already comfortable with for reserving meeting rooms, managing workplace and facilities CAD data, and doing utilization surveys.

Different Sensor Types Zaha Hadid Uses To Understand Client Needs

Arjun Kaicker: We want to get a complete picture of the client’s needs and rely on the following types of sensors to get a full picture:

  • Desk utilization sensors are typically sampling every ten minutes to see if someone is seated at a desk. You can sample more frequently but there is a trade-off between the sampling rate and battery life and most of these are battery powered.
  • Meeting room counting sensors detect not only if the room is occupied but give a count of attendees. This allows us to determine if a room that can hold twenty has someone making a private phone call or is being used for a small meeting with three people.
  • Environmental sensors include noise, C02, and lighting here. It’s interesting to cross reference this with other occupancy data we have. Footfall sensors can measure use of hallways, circulation areas, and breakout spaces.

We can combine different data to answer the following kinds of questions:

  • In a hot desk environment where do people prefer to sit? Which desks typically fill up first and are more frequently occupied?
  • Which hallways are most travelled? How does this affect noise levels?
  • Which breakrooms are most occupied at different times of day?
  • If we have informal breakout areas do the ones that are more isolated tend to get used or the ones near circulation areas? The answers to these questions can vary from company to company and even between departments in the same company. It can be useful to instrument the current space when planning a new one as patterns of space usage intend to continue.

Making Sense of Sensor Data

Yodit Stanton: So far we have been discussing the various types of data that a sensor network can collect on workspace utilization and environment. Let’s talk about two ways that we use most commonly to visualize it to make sense of it. In space: typically as an overlay on CAFM or workspace CAD drawings. This can be used to answer the question “I want to see what’s going on right now.” In time: what are the trends of usage over the course the day or days of the week or weeks of the month. When is peak usage—and perhaps what does this look like on the seating chart? One thing we like to do is incorporate historical data from manual surveys so that we can potentially uncover trends that started before sensors were installed. Data from reservation systems can be incorporated to forecast near term needs and from swipe card systems to cross check total building occupancy.

Practical Cost Considerations For Managing a Sensor Network

Yodit Stanton: I wanted to cover some practical cost considerations for installing and managing a sensor network. You have to look at the following costs:

  • On-site sensor hardware is the most obvious cost and we often try and related everything to a “cost per sensor” or “cost per sensor per month/year” but this is really only a fraction of the total cost.
  • On-site network devices like gateways used to access an Internet Service Provider network.
  • Cloud services include off-site hardware; we say “in the cloud” but it’s in an off-site data center somewhere.
  • On-site installation costs includes labor for site preparation surveys, labor to install the sensors and gateways, and labor to troubleshoot any bringup problems and deliver an operational network on time and with a minimum of disruption to regular work.
  • On-site maintenance costs includes labor to replace failed sensors as well as to replace failed batteries.
  • Network management costs includes software and services to monitor and troubleshoot ny problems end to end in an operational sensor network.

There are a number of trade-offs but the one key point I want to make is that manual maintenance, changing batteries, and installing sensors can be significantly more than the raw cost of the sensor hardware.

  • Selecting “cheaper sensors” that are less reliable and/or less power efficient and therefore have shorter battery life may be much more expensive when analyzed from a total cost of ownership perspective than a more expensive sensor with higher operational life and longer battery life.
  • There are trade-offs between sampling frequency for events (e.g. how often do you check if a desk is occupied) and we normally sample once every ten minutes so that most batteries last 18 months to two years.
  • We have spent a lot of time developing specialized software just for installation management and ongoing network management to make on-site labor hours as productive and error free as possible—and to know that sensor 659 under desk A7 is not working and to understand why.
  • We also spent a lot of time working with manufacturers and doing our own testing and proof of concept designs to verify specifications. We want to offer our clients sensor networks at the lowest total costs per sensor and that means spending a lot of time testing the actual sensors and sourcing the most reliable low power hardware we can.

Need a well-defined strategy for communication about use of sensors

Arjun Kaicker: It’s really important to clearly explain the reasons behind a workplace project. Normally, it’s something simple like to create a better place for people to work. But if you don’t explain it to them, people often assume that it’s about cost cutting, or it’s about downsizing space. It’s about taking stuff away from them as opposed to enhancing the space for them.

In the absence of clear communication, people assume it’s about them. If you don’t explain that the goal is to understand the needs so you can create better spaces some people might assume that you’re trying to check their work performance.

Showing people the results of the data and not just explaining what you’re doing makes a big difference. With sensors, you don’t just have to provide the information to them at the end of the survey. You can actually do it in real time, so people can maybe click on the dashboard and see what findings of the sensors are in real time. And that can often make people feel much more comfortable with the process.

Sean Murphy: We had one question on that, around sensors only painting part of in picture in large organizations where the issues of utilization are more compelling. Know who is using the space is also important, which would seem to work counter to the privacy concerns. But I can understand where architects might want to know which groups or which category of persons.

Arjun Kaicker: Yeah, there definitely is a balance to be struck there. What we do is to have open discussions about with the client about what level of anonymity they want to have. So, there can be complete anonymity or there can be, for instance, anonymity that doesn’t tell you who the individual is. That might provide data on what group they’re in, what team or department they’re in, or might, alternatively, give information on what level they are within the organization. If they’re executive, or if they’re general staff, et cetera. Obviously, if you start to cross reference that a bit too much, then if there’s only one executive in a particular team, that starts to kind of ruin the anonymity.

But generally, you’d be able to process results that anonymous enough and, really, so no one is ever seeing who the individuals are. I think there’s one for the caveats of that, which is that we do … There’s an obvious issue with if there isn’t hot desking, if people have the permanent desk, then you’ll be able to pretty quickly work out, even if it’s anonymous … If that is the only person who ever sits there, then you kind of know how much time they’re spending at their desk. And I think that was always an issue with the space utilization studies. That people have to be comfortable with that level of visibility of what they’re doing.

Closing Thoughts

Sean Murphy: What I’ve learned today is that architects are using data to fuel design and are moving from rough rules of thumb to incorporate more granular data in the way that they’re making decisions. Open Sensors aggregate and help you understand the data. They are moving to enable this information to be fed into their existing tools and existing systems, the cafm systems, the reservations systems at co-working facilities, systems like that.

Arjun Kaicker: I think that sensors are a great additional tool for architects and designers. I don’t think that they provide all the answers for understanding, using these, but they’re a really powerful part of a tool kit. I think, that also just interviews, surveys, workshops with people, really bringing users into the process is still as useful and viable as it’s always been. I also think that what Sensors can start to do is that they can give us more broad data. When we start looking at a series of buildings and how a sense of data might be different in different buildings, and that might be particularly useful for developers even more so than specific building occupiers and so it can really start to help us to understand how to design spaces better for maybe multiple tenants.

Yodit Stanton: As a technologist, it’s very interesting seeing the kind of maturation of the sensor installs and actually enabling people that are not very technical to work with these types of stats. I’m fascinated what kind of impacts these trends are gonna make. Both in terms of the relationship between the levels and occupiers and how the trends that kind of started with, or are starting with, replacing a lot of the manual subways will drive a lot of automation, a lot of a kind of automation with in terms of meeting rooms and so forth and seeing what kind of change it drives in terms of the designer of these spaces. Because, you know, I think everyone wants to, or at least is trying to go towards multi-use, multi-purpose buildings that, you know, we still have some ways to go with that.

Next Generation of Workspaces Event

OpenSensors co-hosted a panel for invited guests on the Future of Workspaces with Cushman & Wakefield. The panel also included Yodit Stanton, CEO of OpenSensors, Uli Blum, Architect at Zaha Hadid and Simon Troup, Founder of Fractalpha. Juliette Morgan, a Partner at Cushman & Wakefield moderated the panel. It was a lively crowd with a sense of urgency – wanting the future now!

Key takeaways

Our panelists gave a view of the current state of data driven workspaces through their different lenses.

Data driven world

For Uli Blum, Architect at Zaha Hadid the world is increasingly driven by data. It gives us much more understanding of the technical aspects of how people work and are living in our spaces. He shared about different work styles, variations of acoustics across a floor, lighting conditions, proxemics, adjacencies, and connectivity. Zaha Hadid wants to better understand all of these aspects and take into account in design.

Competitive edge

Simon Troup, Founder of Fractalpha shared how with data you are trying to find that secret sauce that differentiates you from the competition. He gave an example from the financial market where having access to early data before your competition is a huge edge over them.

IoT traction

Yodit Stanton, CEO at OpenSensors shared about the traction she was seeing, the practical side of how companies are deploying sensors and how to get started. Lots of people are putting in desk meeting room footfall sensors and trying to understand how many people are in the space and how to design better. But we also see combining this workspace occupancy data with facilities data from access control and building management systems for a full view of what is happening.

Say Goodbye to Clipboards! Why Sensors Are Replacing Manual Desk Occupancy Surveys?

Over the past two decades, clipboard reports have been the foundation for desk occupancy studies. In a typical study, 12 undergraduates walk a 5km route through an office workspace to document desk and conference room occupancy. The path takes about an hour to complete and once they finish the start around the path again.

One of the primary benefits of the desk occupancy sensors is that companies can make improvements in how space is used and the potential for reducing costs and energy usage. By capturing and centralizing utilisation information, and doing so in a timely, automatic, non- intrusive manner, analytic programs can find places for improvement.

  • Staffing cost – Manual surveys are expensive, and the biggest expense in such studies is labor. The staff cost is not just for gathering the information, but additional resources are needed to do the reporting on the data.
  • On-going staff training expense – Because of the high turnover rate of these surveyors with clipboards, companies spend a surprisingly high amount of ongoing training and hiring activities. Often this is a very large hidden expense.
  • Errors – Walking a long tedious route gets boring, surveyors make mistakes, and the quality of the study suffers.
  • Sampling rates – Because of the large staff cost, manual surveys are usually constrained to about a week. Sensors enable you to get a better picture of what is going on as you are measuring for a longer period of time i.e. minimum of 8 weeks or permanently. Also, rather than sampling what is going on every hour, you can now sample every 5-10 minutes. The rule of thumb is, you’ve got to sample at twice the event frequency to have confidence in what you’re doing. If you’re doing an hourly survey, you’re really only capturing events that last 90 minutes to 2 hours with any kind of accuracy. On a 10 minute sample, you’re catching stuff that’s 20 minutes, half an hour long. On a 5 minute sample, you’re probably catching events that are 10-15 minutes long.
  • Reporting – The whole point of the study. With manual surveys, whether using pencil and paper or software, staff still need to generate reports. With OpenSensors, the sensors’ data becomes a feed and the reporting and dashboards are ready made and don’t require on-going work to be generated. The whole operation becomes less of a manual process of moving data around; we link with CAFM systems and any other facilities management systems. The process becomes API driven and enables multiple stakeholders to analyse the data.
  • Security – Sensors are less disruptive than having people constantly walking through the office.

Utilisation studies can help you manage desk sharing ratio and unit mix for your flexible working office. Workspace occupancy sensors are replacing manual surveys for a timely, automatic, non- intrusive way to manage wasted desk space and save cost and energy usage.

Why Use Sensors for Workspace Design?

Workspace designers are using OpenSensors’ capabilities to enable their customers to optimise their usage of real estate, smart buildings deliver productivity and improved UX for employees.

Why use sensors for workspace design?

Designers turn to IoT technology and OpenSensors’ digital data layer to address the needs of the owners, facilities managers and building tenants. Innovative new IoT technology and OpenSensors’ data reports, alerts and dashboards provide designers with detailed understanding of how people are using the space vs gut feel on building performance.

A game-changer for the industry

  • Winning more deals both for new development or re-fit of iconic buildings
  • Lower cost than manual surveys
  • Real-time information to facilities managers and even tenants
  • Private data combined with public
  • Understand Air Quality factors for building wellness assessments

Sensors to replace manual work

For the first time deployment and maintenance of smart IoT sensors have become a cheaper alternative to manual occupancy questionnaires and surveys, sensors can have sampling rates of anywhere between once every few seconds to once every 30 mins. This sensor data can be correlated with information from Building Management Systems (BMS) to provide richer context and considerable more insight than manual surveys. Common interfaces include BACnet, KNX and other major systems. These data not only can be combined with private building data but can also be combined with public data like outdoor pollution.

How does it work?

OpenSensors have built hardware, installation and network provider partnerships and relationships to help architectural firms implement smart IoT devices efficiently. We have found that the most successful IoT projects follow a phased implementation approach: Design Phase, Proof of Concept, Pilot, and Deployment. The design phase asks questions such as which sensors, who will be installing and maintaining the sensors. For Proof of Concept, a lab evaluation should include hooking up 5-8 sensors all the way through a gateway to data collection in the cloud. This will give enough real data to verify that the queries and the analytics are feasible. The Pilot Phase ensures that the sensors work at scale and that the gateway configuration has been made easy for the deployment specialists. A pilot phase should be about 40 sensors depending on the density of the sensors. At this point, you can scale up to the number of sensors and the bandwidth required for full deployment.

Practical Examples

Heat maps can help define predictable patterns of usage including peak demand for: * Desks – real-time information of which desks that are in use and which that are available * Conference rooms – Do you have the appropriate amount of meeting rooms, and are they of the right size? * Breakrooms – Where do tenants tend to go and hang out? Are some breakrooms over- or under-utilised? * Corridors and hallways (footfall monitors) – Are some paths through the offices more used than others? Why?

Sensors helps in pitching for new work in a world where people are aware of sensors and how they can drive revenue. Firms who have sensor capabilities have adopted data driven design methods which is replacing gut feel.

Emerging Areas of Practice

Using sensor data enables more accurate planning, and by making it available to occupants, you enable them to both change their behavior and allow them real-time insights and finer customization.

Integration

  • Digital scale models: OpenSensor data can be integrated with architects’ current CAFM systems and 3D rendering environments.
  • Intelligent / Reactive Environments: OpenSensors data can be integrated with displays for open desk notification.

Top 10 Reasons for Data Driven Design

Two of the biggest risks you face as an architect or space planner are:

  • overlooking a key problem in your design or
  • investing too much space or budget for one of the client’s goals, leading to less satisfactory solutions to other goals.

These are often two sides of the same coin: while experience counts for alot it cannot always compensate for a lack of data about how the client is actually using the current space. A small investment in sensors to continually monitor desk usage, hallway traffic, and meeting room and break room occupancy can yield a wealth of hard data to base your design decisions on. This data allows the team to move past dueling hypotheses and get on the same page about real needs based on current usage patterns, which in turn makes the design and development process more efficient and allows the team to craft better solutions for the client’s needs within their space and budget constraints.

The Good Sensor

On a daily basis our customers and community ask us to recommend a sensor provider to buy from, you should ping me on hello@opensensors.io if you want us to recommend your sensor. Often the requirement is vague, “I need an air quality sensor to put on my street for $100?” or “What sensors shall I use to understand my space usage?”. My process of assessment has grown more refined over time because if the sensors we recommend are unsuitable or unusable our company’s reputation is also on the line by association.

So we have come up with our own unscientific way to rate the quality of a sensor that should be applied simply. Most large scale sensor rollout projects of 1K or more often have these requirements as well. It’s possible that sensor providers that don’t rate highly using our criteria produce good sensors but getting the below right takes iteration and discipline in design and the likelihood is that the provider will a higher chance of being able to deliver.

Battery life If a sensor is battery powered, the typical expected life of battery should be clearly stated. Buyers will often want some explanation of what typical means for your sensor i.e. if it’s a PIR sensor have you calculated battery life based on being triggered once a day? The last thing your customers wants to do is invest in a lot of sensors, plus the cost of installation in order to find out that the battery life is only % of what they expected as it will still cost them a lot of money to rip them out and return them.

Bonus point for sensors that publish their battery status as standard so that the sensor owners can have some warning before changing.

Heartbeats

Sensors should tell people whether they are still alive or not periodically. Depending on your battery and connectivity constraints, this can vary, the important thing is that the buyer should not find out a bunch of devices are not working because they haven’t been heard from in days or weeks. Top tip; Heartbeats every 10-60 minutes when possible is sufficient, anymore and it ceases to be informative.

Installation and maintenance procedure

In non consumer environments, the people installing and maintaining sensors are often not the technical design firms or manufacturers. Does your device clearly tell people how to install it, do you have helper applications so that they don’t have to configure firmware? We are working on some solutions for this but more on this later; hint it’s all about enabling people to install sensors efficiently and a non technical installer being able to walk away knowing that the device has joined the network correctly. Does your sensor come with mounting and fittings?

Do people have to unscrew the casing to change batteries? Have you tested this with people and verified it?

 Data Quality

Quality in my definition means, is the data from your sensor easily understandable for someone that doesn’t know your domain. The reality is that often manufacturers pass on the analogue value of the particular sensor and that is too low of an abstraction for most people trying to read it. Battery voltage is a good example, during its life an AA battery will go from 1.5v to about 0.8v, but it follows a curve specific to the device and the battery. Understanding how this maps to a percentage or days of life is often complex. If it’s not possible to do much conversions or processing on your sensor or gateway, perhaps a handy explainer when people buy your device making them understand what the data means.

Support

Please state clear terms for warranties and return procedures to protect your consumers. Consumer protection should naturally apply.

Finally developing high quality hardware is hard, I am always amazed at the skill and dedication it takes when hardware designers and engineers take an idea and get it to manufacturing stage. We try to manage the community’s expectations on sensors they should buy vs the attitude of ‘just throw around cheap sensors’. It would be better in terms of environmental sustainability and user experience to get into the habit of doing more with less sensor density. For more on this, see Dr Boris Adryan’s excellent blog post

I have purposefully not mentioned security in this post as security assessments come with a lot of complexity, will aim to write up on this sometime soon.

Many Thanks to Toby Jaffey for editing.

Tips for Installing a Community Air Quality Sensor Network

Small air pollution sensor technologies have enabled deployment of numerous sensors across a small geographic area that can supplement existing monitoring networks and significantly reduce the cost of longer-term community air pollution studies.This helps mitigate the risk of current approaches to monitoring air quality in a region that rely on only a dozen or so stations and may give you an average that is not be representative of what’s happening where you live.

What are you trying to do

Air quality is affected by many possible contaminants, in fact the Environmental Protection Agency (EPA) has identified six “criteria pollutants” as pollutants of concern because of their impacts on health and the environment . The criteria pollutants (http://www.epa.gov/airquality/urbanair/) are:

  1. ozone (O3) http://www.epa.gov/air/ozonepollution/
  2. particulate matter (PM) http://www.epa.gov/air/particlepollution/
  3. carbon monoxide (CO) http://www.epa.gov/airquality/carbonmonoxide/
  4. nitrogen dioxide (NO2) http://www.epa.gov/air/nitrogenoxides/
  5. sulfur dioxide (SO2) http://www.epa.gov/air/sulfurdioxide/
  6. lead (Pb). http://www.epa.gov/air/lead/

Under the Clean Air Act, the EPA has established primary and secondary National Ambient Air Quality Standards (NAAQS) for these six pollutants. As you begin, keep in mind what you want to measure and how that information will be used. Is there some final output or final report you’ve got to get to?

Understand your sensor choices for collecting air quality data

Commercially available sensors can measure the level of potential contaminants including O3, NO2, NO, SO2, CO, PM2.5 and lead. These devices should be designed to be easy to connect and provide quality data measurements so that non technical community groups can deploy them.

Here are some factors to consider in assessing options for sensors to collect air quality data * cost * operating lifetime * accuracy, precision,and bias of measurement * range of sensitivity * speed of response time * maintenance requirements * reliability

More information on what and how to measure see https://cfpub.epa.gov/si/si_public_file_download.cfm?p_download_id=519616

Beyond the sensors, you will need to make tradeoffs between cost and redundancy for the best network connectivity.

Point to point – lowest cost, greater number of coverage points, least redundancy for each individual point Mesh – higher cost, greater redundancy

Most community-based sensor networks are adopting point-to-point network connectivity because of the ease of connection and low-cost structure. Here is a guide that we already have around pros and cons around connectivity, use that to find the best connectivity network

Our Process

OpenSensors recommends a phased approach, from proof of concept to full-scale deployment, to ensure a successful installation of an IoT network in a business environment. Our aim is to reduce the time to go live and minimize risk.

Phase 1 Evaluate sensors:

Evaluate different sensors for quality, signal-to-noise ratio, power consumption and ease of setup by trying them out on a very small scale in a lab.

Phase 2 Proof of concept:

Do a full end-to-end test to verify that the queries and analytics were feasible. Connect 5 to 10 sensors to a cloud infrastructure.

Phase 3 Pilot phase:

Move out of the lab into your actual environment. Typically, this requires somewhere between 30 to 100 sensors. We suggest a one to two-month test to ensure that the sensors work at scale and the gateway can handle the load, similar to production usage.

In addition to testing the sensors in the wild, this is the time to think through your onboarding process for the devices. Questions like; who will install the sensors feeds into design decisions on the firmware of how much pre-configuration has to be done. We recommend a ‘just works’ approach and an assumption that all sensors will be installed by people who willnot configure firmware. If you need to deploy 200-300 sensors, the installation engineers need to be able to deploy a lot of sensors in a distributed physical environment over a short amount of time. It is much more efficient for your sensors to be pre configured. In these situations, we give usually give people a simple interface to enable them to add meta data such as location and elevation. Sensors should be labelled clearly and details pre-loaded on a cloud platform like OpenSensors before they are deployed so that adding meta data information is a matter of 1-2 steps.

Phase 4 Plan and implement full-scale deployment:

After the pilot phase, there should be enough data to verify network performance and your choices for sensors and connectivity, after which, full deployment can be planned in detail and implemented.

Want to create your own Air Quality project?

The EPA Smart City Air Challenge (https://www.challenge.gov/challenge/smart-city-air-challenge/) is now live. The challenge is trying to help communities figure out how to manage installations of 250 to 500 sensors and make the data public. OpenSensors.io is free to use for community projects working on IoT Open Data projects and will be supporting the EPA’s iniative.

Contact us if you would like assistance on sensor selection, network design, or planning a proof of concept deployment.

Path to Smart Buildings

Whether you are a building manager planning efficient space usage or an architect looking to design state-of-the-art buildings, we have broken down the steps to get you to your desired end goal. IoT planning should start with the business needs, of course, and quickly moves from the component layer all the way up to the application layer. We need to figure out what core data should be gathered and ways to effectively leverage that data. These IoT solutions require an end-to-end or device-to-cloud view.

A Phased implementation approach works best.

We have found that the most successful IoT projects follow a phased implementation approach: Design Phase, Proof of Concept, Pilot, and Deployment. The design phase asks questions such as which sensors, who will be installing and maintaining the sensors. For Proof of Concept, a lab evaluation should include hooking up 5-8 sensors all the way through a gateway to data collection in the cloud. This will give enough real data to verify that the queries and the analytics are feasible. The Pilot Phase ensures that the sensors work at scale and that the gateway configuration has been made easy for the deployment specialists. A pilot phase should be about 40 sensors depending on the density of the sensors. At this point, you can scale up to the number of sensors and the bandwidth required for full deployment.

OpenSensors’ Deployments

We have built hardware, installation and network provider partnerships and relationships to help customers get rollouts live efficiently. Either roll out your own network or we will put you in touch with your local sensor installation specialist to take care of the install and maintenance. We are working with customers and the community to understand what is required at each level for your IoT solution and can ease development and integration issues.

Lessons Learned From First Generation IoT Installations

At first glance,Wi-Fi-based sensors seems like a good choice for a non consumer facing sensor network, however, we have discovered that Wi-Fi has some significant drawbacks.

Access

One of the biggest drawbacks to Wi-Fi enabled sensors in a corporate environment at many of the companies is gaining access. Corporate IT often has valid security concerns of hundreds if not thousands of sensors joining the network and have deployed corporate firewalls that block any access. Often this means that we are not allowed to spin up our own Wi-Fi network in order to have a gateway for a customer’s IoT sensor network. If IT has already deployed a Wi-Fi network they are rarely willing to provide the passwords to allow the IoT network devices and gateways to take advantage of it. Relying on corporate Wi-Fi can make on-site installations and maintenance extremely complex and painful. The whole project becomes dependent on the goodwill of a network administrator for access every time maintenance needs to be performed.

Power

Wi-Fi has good transmission range but that comes with a cost of high power usage. With a short battery life, maintenance costs for Wi-Fi sensors are higher than low-power alternatives. One wireless protocol that is we see in many successful deployments is LoRa because it offers long transmission range at a much lower battery usage than Wi-Fi.

Moving to LoRa and other long range protocols

If you follow our blog and publications, you will notice we have been talking a lot about network technologies, this isn’t a coincidence. We have spent a long time evaluating and piloting these stacks with our community.

Network access and battery constraints are driving the move to long range networks and off WiFi for many IoT installations. LoRa is working well for us so far for a number of use cases most of our customer spin up a private network. The ecosystem of providers is maturing and we are finding a lot of companies who are adopting existing sensors for their networks Gateway providers such as Multi Tech provide good support for the long tail of small scale (> 250 sensor installs) hardware providers to thrive.

LoRa is a wireless protocol that promises between two and five kilometers transmission range between sensors and gateway, if you haven’t already done so please read our introduction to what it is. With a separate LoRa network, facilities and/or operations can install and manage the whole operation without the access and security issues of using the corporate Wi-Fi network. A typical network will have hundreds of sensor devices sending messages to a gateway. The LoRa gateway is a self contained system, we can have the LoRa network sit completely outside of the corporate firewall (GSM) and minimize IT security concerns.

One LoRA gateway can normally cover an entire real estate. This can significantly reduce infrastructure, deployment, and administration costs compared to other shorter range wireless options like Zigbee or Bluetooth that requires complex installs. Our aim is to have a system that non technical engineers can roll out and support, more on how to do this on later blog posts, but in most cases the OpenSensors team is the equivalent of ‘2nd line support’ to the onsite team who have intergrated our apis to their helpdesk ticketing systems etc.

LoRa networks can be public or private. An example of a public network is The Things Network, we continue to work with and support that community. Most current commercial projects are running private networks at this time but will be interesting to see how that evolves over time.

To conclude, LoRa is working well for us at the moment but we will keep researching other networks to enable us to understand the pros and cons of all the network providers. Sigfox is a very interesting offering that we will properly test over the next few months, for example.

Savvy Building Managers Use Sensors to Reduce Operating Expenses

Sensor networks are emerging as a mission critical method for offices and commercial spaces to save money. Offices and commercial spaces are undergoing a smart transformation by connecting and linking HVAC, lighting, environmental sensors, security, and safety equipment. Building and facilities managers are also installing utilization sensors to manage their spaces more efficiently.

Main benefits of data driven buildings * Operational efficiency * Use data for better design * Better workspace experience for employees

Changing workforce

Recently we helped a company design a prototype of a desk sensor monitoring system. Because so many of their people were working from home they wanted to accurately measure the peak demand during the day to see if they could save 10-20% of their desk space. Goals for the system were: * Monitor desk occupancy anonymously. * Minimize installation and deployment costs: rely on solutions that were simple enough that existing non-expert personnel could be trained to deploy. * Minimize day-to-day maintenance and deployment: this drove strategies for long battery life among others. * Design a deployment process that ensured install team could easily add sensor location metadata to allow for rich reporting and analysis once IoT sensor network was operational. * Limit the IT resources needed for deployment

The phased approach works best

First, we looked at many sensors, evaluating quality, signal-to-noise ratio and power consumption. It’s always a good idea to get a handful of different types of sensors and try them out in a very small scale. We chose an infrared red sensor with good battery life-time and a single LoRa gateway that could support all the floors and provide connection to the cloud.

Next we did a full end-to-end test, where we hooked up 5-10 sensors up completely to a cloud infrastructure all the way through the connectivity gateway. Now we had real data flowing into the infrastructure and could verify that the queries and analytics were feasible. This step just makes sure everything works as planned and you will get all the data that you will need.

Once you’re happy with the proof of concept phase, it is time for the real pilot phase. Instead of having just a handful of working sensors, now you’ll hook up an entire floor or a street or whatever your use case might be. It should be somewhere between thirty, forty, maybe up to a hundred sensors. At this point you can ensure that the sensors work at scale and the gateway can handle the load. Typically we see customers running these for a month or two to get a good feel for how the sensors will perform in a production situation.

After the pilot phase, you should have enough data to verify network performance and your choices for sensors and gateways. Now you can plan the full deployment in detail. It’s been our experience, based on a number of customer installations, that the most successful IoT networks follow these steps in a phased implementation approach.

The technology at the silicon, software, and system level continues to evolve rapidly and our aim is to reduce the time to go live and minimise risk. The internet of things is a nebulous term that includes quite a lot of specialised skillsets such as sensor manufacturing, network design, data analysis, etc.

In order to make projects successful, we have taken the approach of building many hardware, installation and network provider partnerships, and relationships to help customers succeed as opposed to trying to do it all ourselves. We have been working with customers to develop methods to lower the sensor density and in turn lower the cost of projects whilst still getting comparable accuracy.

Contact us if you would like assistance on sensor selection, network design, or planning a proof of concept deployment.