The first to be second: Scaling from IoT pilots to mass deployments to sustainability

Enjoyed some very insightful sessions at Georgia Tech Research Institute’s IoT for Sustainability conference July in Atlanta. A lot of thanks to Alain Louchez, Jay Sexton, Jeff Evans for putting together a roster of stellar speakers and panels.

IoT is becoming real

IoT is no longer just on its way, it is here now. Whether it turns out to be Gartner’s estimate of 25 billion connected things by 2021, or Ericsson’s less bullish 18 billion by 2022 - IoT has moved beyond the hype and into practical implementations. In a couple of hallway conversations at the conference, a company shared with me the importance they are giving to IoT, and though they did not want to be the first in the market, they are trying hard to be “the first to be second”. I find that sentiment representative of multiple companies, as IoT pilots demonstrate clear business benefits such as operational efficiency and workforce productivity. 

Pendulum shifting from consumer cycles to enterprise and industrial cycles

IoT is noteworthy due to the fact that - after a boom of more than a decade for consumer-oriented technologies - it reflects a shift towards more enterprise and industry-oriented technologies. Smartphone shipments after growing at double-digit rates for most of the last decade, declined more than 6% in the first quarter of 2019; this came on the heels of a decline of more than 4% in overall smartphone shipments in 2018. The reasons are varied: near-full market penetration, slower upgrade cycles due to limited innovation, increasing prices, and ongoing trade challenges. IoT in concert with Artificial Intelligence (AI), Machine Learning (ML), 5G wireless, Cloud and Edge processing is starting a wave of enterprise and industry innovation and investments in multiple "Smart" domains (Smart buildings, cities, energy, manufacturing, retail, transportation). This will lift the fortunes of recent successes such as Airbnb, Lyft, Uber, WeWork - as well as those of new startups that should materialize in this backdrop. 

IoT pitfalls

? A key question that came up at the conference was how to determine worthwhile IoT pilots, and scale seamlessly from pilot to mass deployment. One of the panelists mentioned their process is to brainstorm ideas that will give them say a 2X to 5X return, and then work backwards to come up with a Minimum Viable Product. This is good advice; the challenge though is how accurate the 2X to 5X return eventually turns out to be. I remember a customer in the buildings sector that was looking to deploy Smart technologies to connect their Access & security, Fire & safety, HVAC, Lighting, Water, and other systems with a range of sensors, actuators, gateway devices, to reduce energy and resource usage. Their calculations showed a 15% reduction in annual energy cost, and based on the costs of IoT equipment and installation labor - break-even would happen within 3 years. What transpired though was that, the annual cost reduction of 15% did not materialize. The reason was the less obvious costs that had not been factored in; one the ongoing price for data cloud processing, analytics, and storage - and the changes in this pricing over time. The second was ongoing operational costs; although IoT reduces energy costs, it adds operating expenses of its own (in areas such as power and cellular connectivity and ongoing labor). 

? Another important point to keep in mind is that, the use of sensors and advanced technologies should not be a reason to “set-and-forget”. For example, in recent well-publicized airline crashes, faulty sensors were erroneously setting the flight trajectory, and it was too late as the pilots realized the problem and tried to wrest back control. Imagine the same type of scenario unfolding in areas such as Access & security, Fire & safety, or Heating & cooling - and the potential liabilities. Enterprises need to make sure to have management systems and “sensors that monitor sensors” to make sure that everything is working soundly, and redundancies are in place. I have not even touched on the issue of security being compromised and rogue IoT devices - which requires an article all on its own.

? IoT will generate reams of data which may be beyond the ability of manual interpretation and control - and thus require deployment of AI and ML technologies. This is another area to proceed cautiously. Though there are good reasons to upgrade along multiple technological fronts in one shot, ML and especially its deep learning variants will bring up unknowns in how set points were calculated, and offer limited abilities to trace back decisions. 

Is IoT synonymous with Sustainability?

Getting back to the main topic of the conference: As we deploy IoT solutions, are we simultaneously also becoming more sustainable? At first look it may seem so. For example, by reducing HVAC energy we are correspondingly also reducing the environmental footprint - right? Let me list a couple of areas though that should give us pause:

? Material impact: There have been increasing concern about the issue of e-waste. People discard electronics for the slightest of incentives; many people who buy a new handset every 18-24 months still have a perfectly fine previous model. The increase in electronics purchases has two effects: one is the increased mining for the materials required and the carbon and energy footprint of production. Second, discarded electronics produce large quantities of e-waste and if not disposed properly, can pollute soil and water supplies. This problem is going to become drastically greater as we consider the billions of sensors that will be purchased and discarded as we deploy IoT.

? Processing impact: As we collect data from billions of sensors and transfer to the Cloud (or Edge data center) - there is the cost of the bandwidth in the form of the fiber optics as well as the lasers, multiplexers, repeaters, amplifiers, detectors, and other electronics employed for the transfer. In addition, once the data gets to the data center it has to be processed by traversing through banks of servers, memory, and storage. As the requirements on these components increase, the size of the data center will have to increase or new data centers will need to be designed - both of which will impact carbon and energy footprint. Data centers already consume as much as 3% of total energy consumed, and 650 billion liters of water (indirectly) in the U.S.; imagine the increase as billions of IoT devices start to come online.

Roy Amara, Stanford University researcher and president of the Institute for the Future, used to say - “We tend to overestimate the effect of a technology in the short run, and underestimate its effect in the long run.” Nowhere is this truer than in the realm of IoT. Initial hype and overestimates of IoT have fortunately calmed down. Now we are in the start of the long run. The trick now is to make sure that we not underestimate IoT’s effects - both positive and negative. 

Feel free to comment, or message me for clarification on any point.