Insight: Google Fi QoS and QoE Implications

Posted by Mae Kowalke on Friday, May 22, 2015 with No comments

Google recently announced its beta Google Fi (aka Project Fi) service, a mobile plan with pricing based on actual data usage. This is a potentially disruptive move in the U.S. where carriers mostly offer "bucket" plans. While on the surface, Google Fi seems like great news for consumers (at least those who get invited and use the Nexus 6 smartphone, the only device on which it's currently supported), the service does present some network performance and user experience issues worth exploring.  



Google Fi's value impact is largely related to decoupling communications (calling, texting) from the 'phone system.' Essentially, Google is building a data-only network and using that to deploy standard communications services using a version of Google Talk paired to a phone number that acts as the caller's address (instead of using an email or Gmail login). 

This new service gives users access to the fastest available network at their locationeither Wi-Fi (first choice) or connection to a partner LTE network (Sprint or T-Mobile). As such, from a performance standpoint calls are essentially VoData (Wi-Fi, LTE, or 3G) rather than relaying on standard handoff or roaming protocols. 

There will certainly be a phase where Google Fi guinea pigs lose calls during handoffs between networks, as those networks have glitches (e.g. a Wi-Fi network Google detects as 'faster' than LTE but in reality is highly congested), and as the software gets developed based on user experience. 

The fact is that moving over "best effort" Wi-Fi networks instead of cellular affects traffic flow priority integrityor, in some cases, the ability to prioritize traffic at all. Many Wi-Fi setups are either subject to no class-of-service distinction, or subject to "non-net-neutrality" practices that can slow down streaming from unknown sources. It may be that Google is counting on 'big pipes,' as they offer with their fiber internet service, but in this case they may be surprised.

Compare this to VoLTE, which will have the highest service priority across each session. Google Fi voice will have to deal with the ultimate multi-vendor, multi-provider HetNet! Therefore, packet loss, latency, delay variation, availability, 911/security, and other aspects of service quality are all going to compromised in any number of possible ways. 

We've all made calls on Skype over LTE/Wi-Fi roaming, and it isn't pretty much of the time. Likewise, Google Talk certainly is not the most reliable communications application on the web; however, the added ability to hop from device to device and maintain the same call is very attractive. (Skype has been promising this since 2012, but so far hasn't delivered.)

We also expect some interesting Google Fi end-user agreements covering liability.

And finally, since everything wrapped into Google Fi will be software-encrypted and data-based instead of using voice channels, the energy efficiency of standard SMS/voicemails will be lost. Expect to recharge often, and for your phone to have a shorter lifecycle. 

What's your take on Google Fi?
Categories: , ,