Do not ask for whom the IoT bell tolls. It tolls for thee. That means you…as in, “ready or not, here I come.” The Internet of Things (IoT) phenomenon (or maybe I should say “tidal wave”) ushers in a new era that will drastically impact Wi-Fi infrastructure and end-users in a variety of key areas. If you’re not ready, you’re toast.
I have previously written that High Density High Throughput (HDHT) amounts to an overall 1% set of corner cases while High Density Low Throughput (HDLT) scenarios are the normal use case. IoT will begin to push HDLT networks to their breaking points very soon, and it will happen without anyone noticing that it has arrived. Did anyone actually notice when BYOD arrived? Not hardly…it was just kablam!…and it was on our doorstep all of a sudden.
At first, administrators will try to keep unnecessary devices off the network, but isn’t that what happened with the BYOD phenomenon? How did that work out? 🙂 Just think of IoT as BYOD 3.0. Yes, that’s right, we’re skipping 2.0 because it’s going to be just that hideous to deal with. Of course, HDHT scenarios will become even more challenging when end users begin adding thousands of new devices to an already-taxed Wi-Fi network practically overnight.
What kind of devices should we be on the lookout for? Well, the list is endless, which is why the industry has given it the moniker “Things”, but some recent developments are:
Wearable and Mobile Tech (watches, pens, Google glass)
Office Items (picture frames, projectors, clocks, TVs, Apple TVs)
Environmental Systems / Thermostats
Smart appliances (refrigerators, freezers, ovens)
Automobiles (Audi, GM, Hyundai, Honda, Mercedes)
Security Devices (Skype-connected cameras)
Organizational Assets (e.g. medical equipment, computers-on-wheels)
Some of the issues facing the Wi-Fi industry as IoT rears its Medusa-like head
IPv6 Transition (required for IoT due to the number of devices)
Connectivity Standards (Wi-Fi Direct? Something new?)
Energy Sources (for billions of sensors)
Über-High-Density (new acronym ÜHD?)
Connectivity (How many APs can take 500+ encrypted connections?)
Client-side Security (device identity, security protocols, mission-critical devices now on the network)
Compatibility (client & infrastructure, 802.11 protocols, security protocols)
Interference (client-side could become overwhelming)
Location Tracking (tens of thousands of devices)
Analytics (for many different device types)
Data Center Overload (too many devices, device-to-device communication)
Redundancy of the Wi-Fi infrastructure will be a must (mission/life critical devices now connect to it)
Compliance (adherence to existing rules becomes more complex, and new regulations likely to surface)
So by this point you’re probably thinking, “Sure Dev….whatever man. This sounds like it’s a little ‘out there’. How can we be sure?“
Constrained By Humans
Until now, the density of Wi-Fi enabled devices in an enterprise has been constrained by humans, e.g.
* How many humans are there in a defined physical space?
* In a specific workspace, are there any mechanisms that allow the humans to be more closely positioned to each other than the ordinary personal space comfort zone? (especially important for VoIP)
* How many devices does each human have, and how many devices are they using simultaneously (on average)?
Wearable technology, such as that made by Samsung, and office items such as the nixplay Wi-Fi Cloud Frame increase the device density on/around humans. . It’s no longer just the human computing devices we have to cope with, but now convenience, control, and even climate related devices are entering the Wi-Fi environment.
Has anyone considered that IoT devices will primarily be secured by PSK? 🙂 Can you imagine the ramifications of someone hacking into a simple client device powered by Spark…
…and suddenly they have the PSK for 10,000 devices across an enterprise that includes mission-critical, life-critical, and security-related devices? That’s unthinkable. Keep in mind that these IoT client devices, for the most part, won’t have integrated firewalls, MDM agents, or 802.1X/EAP supplicants. That gives way to a very real need in the industry for what I generically call (Individual Preshared Key (iPSK) security).
Given that many devices don’t have an eloquent way of receiving a PSK (keypad, integrated web server, etc.), there will be a need for the process to happen dynamically as part of a protocol stack. What about something along the lines of the EAP-PSK rfc or similar? Implementations for such a security mechanism would have to scale to unbelievable proportions for large enterprises, and of course there would be a need for on-premise and cloud-based (likely using RADsec) versions of implementations, not too disimiliar to cloud hosted pbx for business communications.
Imagine if IoT vendors decide to only include Wi-Fi Direct security capabilities (via push-button perhaps) for those tens of thousands of devices. Aren’t we fighting Wi-Fi Direct in the enterprise as it is? Multiply it times 10 or more, and suddenly we’ll be overwhelmed with co-mingling Ad Hoc and Infrastructure Wi-Fi. Does this mean we need to allow Wi-Fi Direct in our enterprise APs? 🙂 Oh the joys of change and progress…
What about IP addresses? Well, instead of the industry screaming about how every user will have three devices, each nabbing an IP address for a period of time, now the number of devices used by, worn by, and referenced by (e.g. clocks, picture frames) individual users will grow by an order of magnitude practically overnight. Hello IPv6 – the protocol most people love to hate will finally have a real driver. No longer will we consider, as part of network design, “on average, each end-user has X devices“, but rather, “there are X devices per unit area.” That will give us valuable information about how to scale the WLAN and how to plan the IP network.
What about location tracking? Already today’s RTLS systems are constrained by the number of devices connected to the Wi-Fi infrastructure, but when multiplying it by at least 10, they will have to change how they think about tracking devices. RTLS is very important technology in healthcare, retail, logistics, manufacturing, and some other vertical markets.
What about frequency space? Because much of the first wave of IoT will be Bluetooth enabled (due to battery life), the devices will be more of an RF interference problem than anything else. That’s certainly a problem given then sheer number of devices that will be in the 2.4GHz band, but it gives us at least some (though minimal) amount of time to prepare for the inevitable move to Wi-Fi enabled IoT devices. The 2.4GHz band is already dirty enough, but now it will be an outright mess. It then behooves network designers to strongly consider moving any mission-critical applications to 5GHz as soon as possible. Already we design enterprise Wi-Fi networks around 5GHz parameters, which leaves us with 2.4GHz messes to clean up, but IoT changes everything.
Have you considered having a corporate policy that says “5GHz only for the corporate and guest networks”? Consider, that taking such a stance would delay the negative impacts (interference, density, security, etc) of IoT for your organization while simultaneously improving performance of your Wi-Fi network (e.g. APs have only one radio’s traffic to deal with). Consider also that single-radio and dual-5GHz-enabled APs could suddenly capture more attention.
Wireless Intrusion Protection Systems (WIPS)
Have you considered that WIPS will get a 2nd, and even bigger, bit at the apple? With so many devices connecting to the Wi-Fi infrastructure, APs won’t be able to do background scanning (because it would fail to service too many clients while doing so). Because IoT devices will primarily be secured with PSK, the WIPS will need to focused on the client devices, and WIPS policies should be created dynamically (instead of statically by an administrator). That much time spent focused on watching huge numbers of clients (some being mobile) will require an overlay WIPS (though not necessarily from a traditional, dedicated overlay WIPS provider). Vendors will begin overlaying their own infrastructures with more sensors of their own (APs acting as dedicated WIPS scanners) so that clients can be monitored and intrusions can be prevented.
What if in your Wi-Fi network design, one radio in some of your two-radio APs is 802.11ac while the other radio is a dedicated dual-band scanner? That could provide cost effective flexibility in your design.
IoT devices will range from the expected environmental items like Wi-Fi & cloud-connected freezers in retail and healthcare to the near-ridiculous. I can imagine executives who are “so busy” that they “need” (cough, cough) to get their news updates while in the shower….
Sure, whatever you say Mr. Super-Busy Executive Guy.
We’ll eventually see sensor networks turning our cities into smart cities, so metro Wi-Fi matters.
Lest you think this is some passing fad, consider the number of companies making big investments on this phenomenon.
If I tried to write everything about IoT into one blog, it would be a book instead of a blog, so I’ll leave a little information for my next installment. In the meantime, here are some nice articles that I’ve read while researching IoT:
I hope this blog got your synapses firing and brightened your day as you started considering the amazing new opportunities in Wi-Fi over the next 10 years due to the IoT nightmare….uh, challenge. 🙂 Keep an eye out for my next IoT installment, and feel free to offer feedback/comments if there are IoT-related topics that you want to hear more about.