Many people refer to the magic that will ensue when our cars’ windshield wipers can talk to weather reporting stations to report the weather. This is often followed by a panicked assertion that we will run out of IP addresses when every windshield wiper has an IP address, or panic about the already overloaded electrical grid going haywire, and assertions that we will need 5G to take us there.
Let’s start with IPv6. Going to IPv6 network addressing and routing protocol had begun to seem like the onset of the metric system in America. When I was in EE grad school in 2000, IPv6 was supposed to be implemented before I graduated. In fact, World IPv6 Implementation Day has passed – twice – in January 2011 and “this time for real” in June 2012. Yes, we are still running IPv4 and IPv6 simultaneously, but the most active domains in the world are running permanent IPv6, including Comcast, Google, Yahoo, Facebook, Yandex, YouTube, Akamai Technologies, Limelight Networks, Microsoft, Vonage, AOL, MapQuest, T-Online, Cisco, Juniper Networks, Huawei, the US Department of Commerce, MasterCard, the BBC, and Telmex (source of list: Wikipedia), as well as many service providers and content providers. IPv6 is the new way to go; there is no reason to field any new service without it, as long as “new service” means “new stack” or “new hardware.” I understand some cloud services (AWS) were still using IPv4 as recently as 2015, and may still be. IPv6 enables 3.4×10^28 IP addresses. That’s not just a large number: it’s 600 quadrillion addresses for each square millimeter on Earth. If you happen to have a car with wipers on the headlights, windshield, and rear window, you can have two sensors on each wiper and we are not going to run out. Granted, part of the problem that led to adoption of v6 is that v4 addresses were allocated by fixed class, and assigned to particular domains. All the v4 top-level address blocks have been assigned, with space reserved for growth in areas of the globe that have fewer addresses now but will need them in the future. At Implementation Day, only 14% of the available 4.3 billion v4 addresses had actually been assigned. Despite the staggering number of v6 addresses available, the Internet Engineering Task Force also built in a more automatic ability to renumber addresses at the host router using self-configuration. We are not going to run out, not in my lifetime, maybe not ever.
The other part that bothers me about assigning IP addresses to a windshield wiper in the first place is that a wiper basically only has one message to send: on, or off. It does not need its own IP address to send what amounts to a 1 bit message payload. The wiper does not need to have its own transmission device at all. Most wipers are going to be communicating with the car, and the car is the unit that sends the more interesting message indicating geo-location and wiper movement that would be important to a receiving weathercaster or weather service. The car may have a whole lot of interesting things to say, such as, external temperature, precipitation, braking action, geo-location and speed of travel. I hope one day the car will be looking for potholes in the road and reporting those too. So it makes sense for the car, or more rightly, the cell connection that the car hosts, to have an IP address (which your cell phone does have.)
Under some visions of the future, the car might be reporting route intent and communicating with a traffic operations center about the costs of different tolled roadways to get to the destination, as is being tried out by ITS in Japan, to mitigate congestion and spread traffic to lesser used routes, typically in commuting. US DoT has a vision that the car or driver may receive signals from an operations center in which the driver is informed that faster or cheaper routes exist if the driver parks the car and takes transit. This is actually being tried in Sweden and in Canada, as apps on a Smart phone. Here’s another point of departure, not so much a myth as what I see as a disconnect: US DoT has achieved an RF spectrum allocation around 5900 MHz (technically 5.9GHz) for short-range car-to-infrastructure (V2I, where v=vehicle) and car-to-car (V2V) communications. DSARC, as I see it, is being developed as everyman’s collision avoidance. It is short range (1000 meters or less), so it’s not very useful for traffic advisories, unless DoT puts an ops center transmitter in every half-mile of road. The DSARC allows cars to sense each other and sense labelled light posts, for example. If cell phones are given an app to emit a DSARC signal, the cars can sense pedestrians and cyclists with cell phones too, but this requires the pedestrians and cyclists to have an active cell on. Perhaps there could be a simplified version of the DSARC fitted to a bracelet like a FitBit, so that small children and pets could have one too.
The centroid of my disconnect is that DSARC can’t do traffic re-routing because of the range limit, unless all data is routed to a central operations center and repeated out. Traffic routing is one of two highly desireable applications of DSARC. However, your cell phone already does a traffic report, so DSARC doing it would be redundant. The bigger difference is that DoT would have to pay for those transmitters and receivers on every half-mile of instrumented road. Alternatively, DoT would not have to pay for all the cell phones and apps in the US – those are essentially “free” to DoT, if it allows itself to harness the power of existing tech. I agree that an “everyman’s” V2V collision avoidance system onboard would be spectacular. I love the automated car future because it brings the possibility of no more highway traffic fatalities, the possibility of saving 33,000 friends, parents, and children per year. But it looks like automated cars and cell phones are going to take us there before DSARC does. Most auto manufacturers are incorporating collision-avoidance sensors in their cars now; any manufacturer who does not will run out of business by insurance premiums on an un-sensored car. Such sensors do not need to go to the web to function; the <0.2 sec deployment takes place without even notifying the driver. [Spoiler alert: I’m going to update my automated car census from 2015, what each car can do and how, in my next blog, 7/24/16.]
Another alert that frequently comes up is that all those instrumented cars and IoT are going to bring cell phone telephony to its knees, or the companion observation that 5G networks and phones will be needed to handle the traffic. It is important that 5G has not been standardized or defined; at present it is an idea that a lot of people are talking about. Dr. Seizo Onoe of NTT Docomo has a pretty hilarious spiel dispelling a lot of printed nonsense on the topic. However, following the typical progression of new generation wireless services, it makes sense the 5G would also skip up the GHz frequency range. At these frequencies, everything is digitized to be stepped up to the high frequency carrier wave, and very compact coding schemes make the most of this range. It is true that there is plenty of capacity in just a little bandwidth at the GHz range. IoT signals are natively digital, so keeping the overhead low will have big impacts, given that most IoT won’t have much to say (e.g., on/off.) For IoT, a home-based network may make more sense, as most consumer-based messages will have little import beyond the home, unless they are mined by permission for marketing purposes (i.e., only you and the food-industrial complex care which mustard you prefer.) Millimeter wave technology, if 5G comes to that, has some interesting multi-path and reflectivity properties that mean potentially better signals in dense environments but a worse experience over longer-distance cell-to-tower environments. Dr. Ashwin Sampath of Qualcomm has given a few presentations on testing results that bear up the faster-more-data-better-sound adverts for 5G, though I note that they potentially come at a cost of more hardware in the phone. So it appears that 5G would improve the user experience and each user can stream more data over the same network. But, more distinct addresses (more “things”) will be a problem solved more by v6 than by 5G.
Finally, returning to the myth of the electric grid being overwhelmed. This is another area of technology (in addition to the vision of no more highway fatalities) that re-assures me that the future for humanity will look more like “Star Trek” and less like “Mad Max:” the incredible shrinking diet of energy for existing devices. Anyone who has replaced all the incandescents in their home with LEDs, kicked the square TV and computing tower to the curb for a thin sheet of glass, replaced the 10+year-old refrigerator, furnace, or airco and watched their utility bills drop has seen this in action. Next step are devices that harvest energy from the ambient atmospherics, whether it is rf waves (see Freevolt, www.getfreevolt.com, currently demonstrating measurement of air quality in London), exotic metals (http://phys.org/news/2016-07-electricity-salt-three-atoms-thick-membrane.html?utm_source=nwletter&utm_medium=email&utm_campaign=daily-nwletter), or harvesting ionic waves. Researchers are looking to power televisions and laptops with these energy sources; a low-power IoT sensor will be easy. IoT sensors will need to run on wireless energy sources, and for low-power transmission, those are to be had. In the near term, inductive charging in your IoT-enabled refrigerator to count the milk cartons and soda bottles left will launch IoT applications. I look forward to the day when my refrigerator notifies my phone when I am low on barbecue sauce, so that I swipe to add it to the shopping list, and watch it delivered to my house Saturday morning by grocery drone.