Wi-Fi is Broken

October 31, 2012

Wi-Fi is broken. It is too technical to use, and no market education will fix it.

Stacey Higginbotham on GigaOm tried to get to the bottom of why the Wi-Fi experience isn’t as good as it could be. She went all technical and ended up with an interesting conclusion:

The problem is all Wi-Fi is not equal, and the industry and providers of Wi-Fi networks have so far done a poor job trying to explain that to the average consumer.

Sure there are technical reasons why Wi-Fi is poor most-places, but guess what: I DON’T CARE!

Here are two ways in which Wi-Fi has been broken for me this week alone.

1. The hotel room case

This last weekend I’ve been on vacation with my extended family – nothing fancy – just went down south for a couple of days.

One of my older relatives has recently moved to the iPhone. For those who are not aware, the iPhone is an optimistic sort of a phone – if it finds Wi-Fi signal somewhere – it signs in to the service.

Only problem was, the hotel we stayed in had a paid Wi-Fi service. So the phone preferred a Wi-Fi pay wall instead of just using the carrier’s data network with the existing paid data package on it. reasonable for home use, but not in this case.

I ended up manually setting this relative’s phone to disable Wi-Fi. I also needed to instruct him on how to enable it again once the vacation was over.

And to think he had to go a full day without internet…

2. The large company case

I work at Amdocs. In the facility I am in there are 5 floors in my building with several hundreds of employees in my floor alone. It is a large place.

We’ve got Wi-Fi all over the building with great reception. The issue is that my Android phone (and I think all phones today) doesn’t switch the Wi-Fi hotspot it is connected to until reception is so bad you can’t even get a trickle of bits on it. The sad thing? My room is located quite far from the hotspot I get connected to when I get into my floor. So I usually have bad Wi-Fi reception in my room – at least until I disable it and re-enable it while sitting in my room.

I move a lot around the building to talk to people and attend meetings. Sometimes, it is simply easier to disable Wi-Fi altogether and just use the cellular data network instead – you don’t get flaky reception in the same scale you do on Wi-Fi when you know there are close hotspots nearby.

Why can’t Wi-Fi just switch to the best network automatically?

It is broken in both cases because there’s no way to manage it properly – no spec or best practices that work well enough.

As I said – it is time for Wi-Fi to go mobile. Make Wi-Fi fade into the background and don’t have me manage it manually.

You may also like

Two years of WebRTC Insights

Two years of WebRTC Insights

Your email address will not be published. Required fields are marked

    1. Is it really that complicated for a phone to connect to a wifi network and send some packets to check if it is actually connected to the real Internet before it prefers to use it over any other network ?

    2. There’s a solution using an app for the other problem as well – but why the hell do I need to find and install an app for something that needs to come out of the box?

      How many people suffer from this problem and don’t even think of looking for an app to solve it?

  1. From the top of my head we basically have a number of problems (from reading the article and your post):
    – 1. Maybe Wi-Fi NICs can’t connect to two access points at the same time to get a real sense of how fast they can talk to the access point ? Or maybe the problem is that the higher layer does not have access to an API to request it ?

    – 2. address portibilty: when you connect to a different Wi-Fi Access point you have no garentee you’ll still have the same address. Only when it advertises the same network you have some chance.

    – 3. TCP, TCP can’t handle huge delays from bufferbloat (the backhaul mentioned in the article) and Wi-Fi can add huge delays itself everytime there is interference or congestion like the crowded places the article talks about.

    It sounds to me like all they could all be solved.

    1. Maybe we need an API for Wi-Fi NICs and to make the OS smarter ?

    2. Multipath TCP can solve the address portabilty problem: http://www.youtube.com/watch?v=02nBaaIoFWU

    3. Bufferbloat at the backhaul can be solved and maybe TCP can be made smarter too, just look at Coded-TCP (they claim they can recognise the difference between interference and congestion):

    So if someone is willing to invest time and money in operating system development and deployment (ha ! there is a problem !), we might actually be able to solve it.

    If only client devices like smartphones deploy Coded-TCP and also multipath TCP and CDN’s deploy multipath TCP it would already be a huge improvement. Most things might even go away already.

  2. This is a two-way street. Both the devices and infrastructure should get smarter. But I can tell you that in US, my general experience is quite different.

    For your office, if your IT organization would install Xirrus Wi-Fi antenna arrays, I guarantee that you would have great experience anywhere in your building.

    Overall, in US, the Wi-Fi infrastructure is being built up very significantly, to a large degree thanks to the cable companies competition with the phone companies. I can drive around Stamford with my wi-fi iPad, and it will pick up a signal and connect to the CableVision wi-fi pretty well and automatically.

    I can tell you, if not for the wi-fi support in my Blackberry (and thanks to T-Mobile UMA), I wouldn’t be able to use my mobile phone as we had cellular service down for two days already as an aftermath of the hurricane…

    All in all, the state of Wi-Fi appears to be pretty decent in US.

  3. What about having some built-in connection manager w/ some intelligence to just ping the Wi-Fi network and get a sense of how busy it is – if there is a lot of latency, then the freebie Wi-Fi is no good. Similarly if it’s a paid Wi-Fi, then check to see how much credit is left on the macro cell network connection (if unlimited, all-you-can eat plan, then this becomes a superfluous question, but those are rarer; if there is a cap after which the user gets throttled down, if close to the cap, the connection manager can ask a Y/N question to the user). But that’s a bit of a dream.

    A simpler step 1 would be for this connection manager to follow what I like to call the ABC principle (Always Best Connected). It should be able to detect Wi-Fi vs. macro network SNR and seek the better connection, if economically (and practically feasible).

    But the solution can be further enhanced by some Wi-Fi hotspot analytics and a way for the connection manager to query those analytics. There are a few companies in this space partnering already with connection mangers (e.g. Tutela w/ Green Packet).

{"email":"Email address invalid","url":"Website address invalid","required":"Required field missing"}