Sunday, November 03, 2013

Alfidi Capital Visits ARM TechCon 2013 for IoT, Wearables, and Fun

I visited this month's ARM TechCon 2013 conference down in Santa Clara.  This is where we nerds go when we're ready to graduate from software to hardware.  I suffer from a serious case of engineer envy, because the majority of this stuff is over my head and I really need to understand its implications.  I had never even heard of ARM before this event.

The first morning keynote I attended made it clear that success in building IoT devices would require teams with skills in engineering, programming, and design.  There's no immediate need for anyone who studied business or the humanities.  My conversation with an engineer friend afterwards left me humbled by my lack of understanding of dynamic IP, middleware, and other tech concepts that are much more relevant than my knowledge of finance.  The good news is that I don't have to go back to school for another degree.  The OpenCourseWare Consortium exists to fill my knowledge gaps in STEM fields.


The ARM dude who keynoted was optimistic about embedded sensors in transportation and household goods.  My experience with military logistics exposed me to RFID tags in the mid-'90s but they could only track the origin and final destination of an entire shipment.  Sensors that can relay the ambient temperature of perishable commodities and run routine diagnostics on the prime mover's drive train are the next step that the US military has yet to take.

I hit a couple of product pitch seminars for whatever free generalist views I could gather.  I had wondered whether wearables would be viable without a smartphone, and it turns out the smartphone is they key to everything.  Your phone is the transmission base station that will connect your Google Glass and smartwatch to the world.  I still think battery life and power management are going to be a huge hurdle for wearables.  I'm totally convinced that the first startup to come out with a single-person energy microgrid app that lets a smartphone manage the power budgets of multiple wearables will be a hit.  The Qi standard for wireless power charging from the Wireless Power Consortium is emerging but its short range limits wearables to charging off the battery in a smartphone.  If wearables are going to debut at CES then that's my excuse to go to Las Vegas next year.  I got a very strong impression from listening to product pitches at ARM TechCon that lithium-based chemistry will be the dominant approach in wearables' batteries.  If demand for lithium explodes, it will force the world's three leading lithium producers to revise their estimates for lithium demand upwards.  Basing their projections on demand from hybrid-electric car drive trains will underestimate demand if they're not anticipating IoT!  Oh, BTW, Afghanistan has plenty of untouched lithium deposits.  Now you know why the US will never completely withdraw from that country.

The only exhibitor on the expo floor I found noteworthy was ChipEstimate.  I like that they can estimate a chip's cost and energy use from its design.  I hope it works for Arduino tech.  I'll bet a bunch of members of the ARM Connected Community will need their chip designs tested once they license ARM's IP.

I used to read about the term "middleware" in dot-com era computing periodicals but I haven't heard it since then.  Middleware may be making a comeback.  If IoT device makers are going to tout connectivity over stand-alone capability, it begs the question of how the device will relate to the network.  The only practical lesson in connectivity I recognized for IoT projects was the need for the 6LoWPAN stack to replace the TCP/IP stack, because the new stack standard must be compatible with IPv6.  One of my pet peeves about conferences is a presenter who posts incorrect URLs for sources they cite.  Some IoT presenters cited standards bodies for their stack protocols that I have not been able to locate with the URLs they specified.  Come on, techies.  You people are engineers and I'm providing a public service by pushing your knowledge to the finance community.  If you can't get your source correct, don't bother citing it.

I wanted to hear about cloud access for the IoT.  Product pitches for smarthome devices aren't good enough.  I say that smartphones as the gateway to the IoT-enabled home pose security risks if they are stolen.  Multi-factor authentication that includes biometrics will ensure a thief can't use your stolen smartphone to enter your smarthome by posing as you.  I also keep hearing that data mining will reside in the cloud's enterprise apps but IoT devices will operate in the "fog" at the edge of the cloud's network connectivity.  That obviously means the ERP cloud does not extend to whatever drives and IoT devices must connect.  The ERP system and the device must connect through some gateway, which I take to be a stalking horse for middleware.  If the smartphone is part of this gateway, it must have the authentication protocols I described above.

Dr. Daniel Kraft gave the Wednesday keynote.  I've met this guy on a few occasions but this was the first time I got to hear him share knowledge.  The dude is brilliant.  His observations came at warp speed from many directions as he integrated the cloud / mobile / Big Data trend into health care diagnosis and treatment.  He nailed the problem with his revelation that medical technology has been slow to adopt digital connectivity.  This leads to case-by-case practices differing by geography for the same treatments, with no broad applications of best practices.  I did not know that disruptions in sectors happen at the "knee" of an exponential growth curve where many technologies converge.  Dr. Kraft cited multiple examples of technology that will make more data available at lower price points, with cloud-based analysis that will lower treatment costs.  He provided a live case study in the quantified self by streaming his own real-time data from several wearables into his presentation.  Dr. Kraft's adaptation of aviation concepts like checklists, heads-up displays, and simulations now apply to reducing errors in medicine.  His talk was an excellent preview of the material you'll find at Singularity University's FutureMed 2020 conference.  Check out Uncle Sam's Health Data portal for metrics and concepts that will be very relevant to health sector entrepreneurs.  I look forward to seeing "data donors" become as prevalent as blood, bone marrow, and organ donors.  BTW, the optics of Dr. Kraft's presentation were mind-blowing.  This was nothing like a typical PowerPoint lecture.  I have got to learn how he integrated live data feeds into a slide show that could zoom, pan, fade, and display animation.

Wednesday's closing chat with ARM's CEO featured the very attractive Michal Lev-Ram.  She can chat with me anytime, if you know what I mean.  I get the argument that low-cost, low-power sensors will appeal to energy austere markets but IMHO they also need to function in environments that have austere connectivity.  I don't know whether the rest of the developed world is ready for IPv6.  If enterprises are going to effectively monitor their on-site usage of physical resources, as ARM wants, their utility providers need to be early adopters of the new connectivity standards.  I really think that will happen in energy-rich markets first where utility grids are the most robust.  The developed world will have to leapfrog its connectivity in several realms simultaneously - Internet, electricity, and transportation - for IoT to see the explosive growth its advocates want.  Rapidly falling smartphone prices have also convinced me that growth in the mobile sector still has some life as late adopters come in at lower price points.  Talks like this one at the ARM conference have further confirmed my theory that business leaders' skill development will have to change.  I now think that IoT is a major opportunity for the fusion of knowledge management, operations research, project management, and Six Sigma into the integrated teams that will optimize whatever the IoT universe feeds into Big Data.

The Thursday morning keynotes taught me that cellular connections provide access for low-power sensors.  I hope these aspiring IoT device makers have cleared that with telecom carriers and obtained their FCC certifications before they begin installation into downstream products.  I've noticed more tech thought leaders are now claiming that Small Data will be the foundation of Big Data.  Small Data from low-power IoT devices and sensors seems like a good early candidate.  Standards bodies like IETF, ZigBee, Open Mobile Alliance, IPSO Alliance, and One M2M are proliferating.  I sure hope they talk to each other to avoid duplication of effort.  I also hope they address the middleware layer of connectivity between devices and networks, where interoperability is apparently not as common as interoperability among devices.  My sense of the space between IoT devices that generate Small Data and the ERP back ends that generate Big Data is that middleware connects the two realms.  Heck, I don't know if this makes sense to an IT expert because I don't speak an IT language.  I'm going to track the use of the term "middleware" at future conferences to see if vendors are pitching solutions that connect these Small-to-Big gateways.

I checked out a session on kickstarting IoT device development.  I am now more certain than ever that the basics of manufacturing are a lost art to a generation of innovators.  Young engineers are in dire need of crash lessons in manufacturing and quality control.  Small scale manufacturing is not a typical waterfall model of design to production because the process will require multiple physical adjustments after product testing.  Form factors matter as much as connectivity.  Arduino controllers are Web-enabled but physical design specifics may interfere with Web connectivity if they're incompatible with standards.  Pay attention, you aspiring makers reading my blog.  Now you know why I mentioned all of those standards organizations in the paragraph above.  Documenting a prototype's assembly stages is another lost art Americans will have to recover because that's how the makers in this session identified imperfections in their injection molding process.

Outsourcing manufacturing for small-run items means the manufacturer will often charge by the time spent producing units.  Special aesthetic features add time and cost.  A mirror finish on a plastic surface requires slow cooling after molding.  This means the unit cost is higher and the final retail price must be higher.  The presenters discovered that the manufacturing process does indeed influence final product positioning in the market.  What you thought would be a cheap device now becomes a premium product if you adjust the production process to add attractive qualities.  The final bill of materials (BOM) will be very different from your original estimate, and it will affect any revisions to financial projections you present to investors in later capital raising rounds.

The manufacturing lessons make me think of how IoT products will be serviced and replaced if they break down.  Career fields that repair expensive goods like automobiles and HVAC units exist because the replacement cost of those goods is very high.  Replacing a low-cost IoT device should ideally be done simply and cheaply by the user but IMHO that won't be possible if manufacturing adds unanticipated costs.  There is such a thing as cell phone repair but it seems to be focused on recovering data that will transfer to a replacement device, with little concern for returning the device to working order.

Hardware and software integration during prototyping requires iterative experimentation with flexible manufacturing and adaptive APIs.  This approach is well within the expertise of computer and mobile device manufacturers but I don't know how deeply this proliferates within other capital goods producers.  Makers are going to be heavy users of Github if they want to build IoT devices.  Fortunately, there's no shortage of helpful models.  The Mbed ecosystem allows prototyping of ARM microcontrollers.  The open source hardware movement encourages makers to play with BeagleBoard and Raspberry Pi components to their heart's content.

My finance background is of little use in comprehending IoT developments.  I need to get busy reviewing any extant Gartner and Forrester coverage of IoT.  I should also probably read Clayton Christensen's The Innovator's Dilemma for some original disruptive thinking.  One participant at this conference relayed a quote from an anonymous author:  "An idea without execution is hallucination."  I definitely got a few ideas at  ARM TechCon but I'm not yet skilled enough to execute them.  That comes after I do some more homework.