International LTE Data and VoLTE Roaming - NTT Docomo

Quick recap of the Bearer Architecture: Remember the interface between S-GW and P-GW is known as S5/S8. S5 in case the S-GW and P-GW are part of the same network (non-roaming case) and S8 in case where P-GW belongs to another network than S-GW (roaming case). The S5/S8 interfaces are generally exactly the same. There is a possibility of different types of S5/S8 interfaces like GTP based and PMIP based but lets not discuss that here.

NTT Docomo published an excellent article in their magazine recently showing the different approaches to International Data roaming.

The different scenarios above are based on the guidelines provided in GSMA PRD IR.88. Each operator has to adopt one of the scenarios above, NTT Docomo has selected scenario 4. The Home PLMN (HPLMN) and the Visited PLMN (VPLMN) connect via IP eXchange (IPX).

As can be seen above, the MME in VPLMN communicates with HSS in HPLMN using Diameter Edge Agent (DEA).

Finally, it is well known that NTT Docomo is not launching VoLTE untill 2015. The above is their proposal on how they handle VoLTE while in Japan and when roaming.

The paper is an interesting read, embedded below:

NTT Docomo: Implementing LTE International Data Roaming from Zahid Ghadialy

Another article worth a read is the VoLTE roaming with RAVEL here.

VoLTE Roaming with RAVEL (Roaming Architecture for Voice over IMS with Local Breakout)

Voice over LTE or VoLTE has many problems to solve. One of the issues that did not have a clear solution initially was Roaming. iBasis has a whitepaper on this topic here, from which the above picture is taken. The following is what is said above:

The routing of international calls has always been a problem for mobile operators. All too often the answer—particularly in the case of ‘tromboning’ calls all the way back to the home network—has been inelegant and costly. LTE data sessions can be broken out locally, negating the need for convoluted routing solutions. But in a VoIMS environment all of the intelligence that decides how to route the call resides in the home network, meaning that the call still has to be routed back.

The industry’s solution to this issue is Roaming Architecture for Voice over LTE with Local Breakout (RAVEL). Currently in the midst of standardisation at 3GPP, RAVEL is intended to enable the home network to decide, where appropriate, for the VoIMS call to be broken out locally. 

Three quarters of respondents to the survey said they support an industry-wide move to RAVEL for VoLTE roaming. This is emphatic in its enthusiasm but 25 per cent remains a significant share of respondents still to be convinced. Just over half of respondents said they plan to support VoIMS for LTE roaming using the RAVEL architecture, while 12.3 per cent said they would support it, but not using RAVEL.

Until RAVEL is available, 27.4 per cent of respondents said they plan to use home-routing for all VoLTE traffic, while just under one fifth said they would use a non-standard VoLTE roaming solution.

Well, the solution was standardised in 3GPP Release-11. NTT Docomo has an excellent whitepaper (embedded below) explaining the issue and the proposed solution.

In 3GPP Release 11, the VoLTE roaming and interconnection architecture was standardized in cooperation with the GSMA Association. The new architecture is able to implement voice call charging in the same way as circuit-switched voice roaming and interconnection models by routing both C-Plane messages and voice data on the same path. This was not possible with the earlier VoLTE roaming and interconnection architecture.

Anyway, here is the complete whitepaper

NTT Docomo: VoLTE Roaming and Interconnection Standard Technology from Zahid Ghadialy

ANDSF: Evolution and Roaming with Hotspot 2.0

Access Network Discovery and Selection Function (ANDSF) is still evolving and with the introduction of Hotspot 2.0 (HS 2), there is a good possibility to provide seamless roaming from Cellular to Wi-Fi, Wi-Fi to Wi-Fi and Wi-Fi to Cellular.

There is a good paper (not very recent) by Alcatel-Lucent and BT that explains these roaming scenarios and other ANDSF policies related information very well. Its embedded below:

Friday rant: OTT, Viber, Roaming, etc.

The same old story, mobile operators are seeing that their revenue is not growing, even though they are upgrading their networks and introducing new features / technologies. The following is from Total Telecom:

The global telecom services market generated revenue of €1.12 trillion in 2012, although at 2.7% growth was slower than in the previous year, according to the 2013 DigiWorld Yearbook published by IDATE on Thursday.

The "DigiWorld" as a whole - which also includes telecoms hardware, software and computer services, computer hardware, TV services, consumer electronics and Internet services – recorded revenues of €3.17 trillion last year, up 2.8% on 2011. By 2016 that figure will have risen to €3.66 trillion, IDATE predicts, with telecoms services contributing €1.25 trillion (see chart).

Telecoms operators are experiencing flat growth, while over-the-top (OTT) providers are seeing revenues increase by 15% a year, Vincent Bonneau, head of IDATE's Internet business unit, told attendees at the DigiWorld Yearbook launch in London earlier this month.

Another interesting piece of news was that Viber has launched a desktop application which means it can now rival Skype fully.

Guess what, I would think that operators have more to worry from this news than Skype. I have stopped using Skype for some time now due to many issues I have with it and have moved to Viber for a few months.   If you are a regular reader to this blog then you would have read my recent post complaining about the global roaming rates. When I am travelling abroad, I make sure there is WiFi and use Viber as a substitute for Voice and SMS. In fact I can send MMS and emoticons using Viber which would cost a fortune over cellular otherwise.

Sometimes it feels like the operators are sleepwalking into their own destruction by not innovating enough and fast to be a challenge for these OTT services. Not entirely sure what the solutions are but there are quite a few ideas around to start thinking in that direction. An interesting presentation by Dean Bubley I posted here is a good starting point. Another one from him and Martin Geddes is embedded below, which is quite interesting and intutive.

Enough of my rants, what do you think about this?

Is the Global Mobile Roaming model broken?

Yesterday, I noticed some heavyweights discussing roaming prices on Twitter. It is embedded below using the new Twitter embed feature:

I really don't understand why Google Maps still failsin this way twitter.com/BenedictEvans/…
— Benedict Evans (@BenedictEvans) May 18, 2013

@benedictevans it might be the Vodafone connectivity on your phone. Try O2 :)
— Carlos Domingo (@carlosdomingo) May 18, 2013

@carlosdomingo you'll have to match their roaming pricing first ;)
— Benedict Evans (@BenedictEvans) May 18, 2013

@benedictevans I am sure @ronandunneo2 will be interested in looking at it.
— Carlos Domingo (@carlosdomingo) May 18, 2013

@carlosdomingo @benedictevans ours is a fixed rate of just £1.99 a day for data and #TUGo for IP calling to anyone
— Ronan Dunne (@ronandunneo2) May 18, 2013

@ronandunneo2 @carlosdomingo not in New York, as far as I can see
— Benedict Evans (@BenedictEvans) May 18, 2013

@benedictevans @ronandunneo2 Hi Benedict, you can find our latest rates by selecting USA from the drop down list here j.mp/ML2Xzy
— O2 in the UK (@O2) May 18, 2013

@o2 @ronandunneo2 O2 in the USA: £6 per meg. Vodafone in the USA: £25 for 100 meg a day. Sorry, but that's an easy choice.
— Benedict Evans (@BenedictEvans) May 18, 2013

Those who follow me on Twitter may have noticed me ranting about the roaming prices recently so I thought that this is a perfect opportunity to put my thoughts down.

As being discussed above, I went on the websites of two UK operators and found out about their roaming rates to India and The USA and they are as follows:

 It should be noted that there is a better rate available with some kind of bundle opt-in from both the operators and I have not shown about the other UK operators but they offer a similar sort of rate so I am not trying to single out O2 and/or Vodafone.

Since LTE is 'All-IP' network my interest is more from Data point of view rather than the voice point of view. A colleague who went to India recently decided that enough is enough and he bought a SIM in India locally. Apparently is just a bit too difficult to get SIM in India if you are not an Indian resident, nevertheless he somehow managed it. The rates as shown below was INR 24 for 100 MB of data.

Rs. 24 is something like $0.50 or £0.35. You see my problem regarding the data rates? People may be quick to point out here that India has the cheapest data rates in the world. On the other hand we look at US, the rates are as follows:

Even if we assume $15 / 1GB data, its far cheaper than the roaming rate which may be something like,  £3/MB = £3000/GB or £6/MB = £6000/GB.

I blogged about all the interesting developments that have been happening in LTE World Summit regarding the roaming solutions but what is the point of having all these solutions if the operators cant work out a way to reduce these costs. Or is it that they do not want to reduce these costs as they are a good source of income?

The operators complain that the OTT services are taking business away from them and turning them into dumb data pipes but to a lot of extent its their fault. People like me who travel often dont want to spend loads of cash on data and have worked out a way around it. Most of the places I visit have WiFi, most of my work is not urgent enough and I can wait till I am in a WiFi coverage area. In some parts of the world, still I have to buy an expensive WiFi access but compared to the roaming rates, its still cheap so I have stopped complaining about it. My decision to book a hotel depends of reviews, free breakfast and free WiFi. Some of our clients who give us their phone to use abroad strictly inform us that data should not be turned on unless its a matter of life and death.

If the operators dont change their strategies and work out a better solution for the roaming rates I am afraid that their short term gains will only lead to long term pains.

Do you have an opinion? I am interested in hearing.

VoLTE - making it work

Interesting presentation from Acme Packet in the LTE World Summit.

On LTE Roaming ...

The IP eXchange (IPX) is used for data when the users roam between different networks. GPRS Roaming eXchange (GRX) is a service within IPX. One of the main areas of discussion within the LTE World Summit 2012 in the Signalling Focus day was roaming on LTE. Different vendors have different proposals and solutions; couple of them are as follows:

Interesting to see that iBasis has proposed LTE Signalling eXchange (LSX) as a way forward.

A presentation from Acme Packet (for an earlier conference) has interesting VoLTE roaming options proposal.

Finally, while everyone was focussing on LTE-LTE roaming, only Diametriq was looking at LTE-LTE/3G/2G Roaming. The relevant part of their presentation is embedded below.

Diameter and LTE Roaming

View more presentations from Zahid Ghadialy

Happy to hear more on this topic if anyone else wants to contribute. Please feel free to add comments.

Roaming with the IP eXchange (IPX)

From Wikipedia:

Traditionally, voice traffic interconnection between different operators has utilized the international SS7/TDM networks. However, lately the all-IP paradigm with VoIP is being rapidly introduced by different operators in various forms, such as IMS. In order to minimize the number of conversions between packet-switched voice and circuit-switched voice there is a clear need to deploy an IP based NNI (Network-to-Network Interface) and therefore an IP based interconnection network.

It is also evident that a large number of IP based services (such as Presence or IM) simply cannot be interconnected using a SS7/TDM network, further increasing the need for evolution into an IP based interconnection network.

Since the year 2000 GSM operators have been using GRX (GPRS Roaming Exchange) network for routing the IP based commercial roaming traffic between visited and home operators. Mainly 2.5G and 3G data roaming has been using GRX. GRX is a private IP network (separated from internet) consisting of multiple different GRX carriers that are connected to each other via peering points. However, GRX is limited only to GSM operator community and not all GRX's are capable of meeting the demands of real-time services.

Even though the GRX environment is not entirely suitable as a common IP network for interconnection and roaming, it offers a good starting point for the development of IPX. IPX development has been done in various GSM Association projects and working groups since 2004.

The following presentation is from LTE World Summit:

The IP Interconnect in VoLTE - Market & Technical requirements

View more presentations from Zahid Ghadialy

GPRS Roaming eXchange (GRX) for LTE/EPS Networks

The GSM Association (GSMA) has came to the realization that GPRS roaming based on bilateral relationships between individual GPRS operators is incredibly complex and expensive to maintain, in particular if the number of roaming partners is high. In fact, each operator will have to have N(N - 1) dedicated links to other operators (given that N is the global numbers of operators for which roaming should be supported). The GSMA has therefore recommended the use of a GPRS Roaming eXchange (GRX) for the Inter-PLMN GPRS roaming scenario.

The GRX is built on a private or public IP backbone and transports GPRS roaming traffic via the GTP between the visited and the home PLMN (Figure above). A GRX service provider has a network consisting of a set of routers and the links connecting to the GPRS networks. Moreover, the GRX network will have links connecting to other GRX nodes to support GRX peering between networks.

The GRX service provider acts as a hub, therefore allowing a GPRS operator to interconnect with each roaming partner without the need for any dedicated connections. This allows faster implementation of new roaming relations, faster time to market for new operators, and better scalability since an operator can start with low-capacity connections to the GRX and upgrade them depending on the bandwidth and quality requirements of the traffic. Other benefits of GRX are as follows:

Support of QoS: This aspect that will be very important for the GPRS services and, in particular, for the transition to 3G systems.

Security: The interconnection between the home operator and the visited operator uses the private GRX networks, hence does not require the overhead of maintaining expensive IPSEC tunnels over the public Internet.

DNS support: Through GRX it is possible to support a worldwide ".gprs" DNS root, where the various GRX operators will collaborate in managing the root and each operator's DNS servers will be connected to such roots to provide translation of DNS names specific to one operator.

In conclusion, GRX is introduced for GPRS roaming to facilitate the network operators for the interconnection between networks to support roaming and will play a very important role for the transition to third-generation systems.

In the LTE World Summit, Alex Sinclair, Chief Technology Officer, GSMA mentioned about the important role GRX will play in the LTE networks. The figure below are his views on GRX.

Diagram and Initial text Reference: IP in Wireless Networks By Basavaraj Patil, et al.

More information on GRX is available in GSM Association IR.34 document.