If you've been following my blog for a while you've probably noticed my support for Voice over LTE via GAN, or VOLGA for short. In case not and you are interested here's a link to a short intro and here a link to a whitepaper which I've recently written on the subject. It looks like the VOLGA forum has not been idle over summer and has continued working on the final stage 3 specification that contains all the nitty gritty details on messages, information elements, etc. Now in September 2009, the final version has been published which means all documents are together now to implement VOLGA. For the details, see here. Well done, I am looking forward to seeing the first implementations!
A main feature that is missing in LTE so far is a viable standardized system for voice calls for the short and medium term. Sure, 3GPP has specified CS fallback and many network operators are hoping for IMS in the long-term. However, in my opinion, both leave LTE very exposed on the voice front at least over the next couple of years.
That's why I am a fan of VOLGA, short for Voice over LTE via GAN as it offers a smooth way of connecting the already existing voice infrastructure to LTE. I've already voiced my thoughts on it here and here, but blog posts can't really go into the technical details. So with the kind support of Kineto Wireless, I've assembled a whitepaper about VOLGA to describe how it works and its advantages and disadvantages compared to other voice over IP systems for wireless networks. So if you have a bit of time and are interested in the details, have a look and enjoy!
In a discussion around VoIP someone recently said to me that he thinks that "non-network operator supplied" VoIP is having a big impact on voice prices, both in fixed and mobile networks. I am not quite sure this is yet the case, however. When looking at prices for fixed and mobile voice calls, it can be observed in most countries where telecommunication was liberalized a decade or so ago, that prices are falling. This has consistently been happening over that time, long before VoIP came up.
From my point of view, falling fixed and mobile voice prices are more a result of competition between incumbent and startup telecom companies, who are offering voice services over circuit switched technology.
Technology has advanced, so in many cases, voice calls are transmitted over IP in the backbone networks of network operators and also internationally, but that's still virtual circuit switching and not 'end to end' VoIP over the Internet. Sure, there are services such as Skype, which are free while the call stays between two users of the same service but all people I know use it as additionally to circuit switched services rather than a replacement.
So why are VoIP services still lacking popularity today? I think it's quality of service and ease of use, in which end-to-end VoIP is still very much behind traditional circuit switched voice. That doesn't mean VoIP isn't catching up, but in my opinion we are not yet at a point where VoIP is a serious threat to circuit switched fixed and wireless voice.
As always, comments are welcome.
The other day, a reader asked whether it is true that a voice call over a UMTS circuit switched bearer is less expensive than over a packet switched UMTS bearer. Good question and I guess very difficult to answer as there are many parameters. But nevertheless, let's expand the question and put GSM and LTE on top.
In the GSM world things were simple at first. There's a 200 kHz carrier and you can squeeze 8 timeslots into it. On the main carrier of a cell 6 out of those 8 timeslots can be used for voice, on all others, all timeslots can carry one voice call. Further, the adjacent carrier can't be used due to overlap, so the carriers bandwith is effectively 400 kHz. To increase the number of calls, the network operator can use AMR half rate, theoretically doubling voice capacity. Here it starts to get difficult as a half rate channel should not be used under weak signal conditions, i.e. some calls should fall back to a full rate channel so more redundancy and error correction information can be added to prevent the call from dropping. Anyway, a full rate channel voice coded streams at 12 kbit/s in each direction. Add error detection and correction bits and you end up with around 28 kbit/s.
UMTS Circuit Switched
In terms of resource use, things are similar as in GSM. The AMR full rate codec streams at around 12 kbit/s and redundancy information is added. I'd say resource use is similar as in GSM.
UMTS / HSPA Packet Switched
Packet switched means Voice over IP. Here, things start to get difficult because what is VoIP in practice? There's no standard solution as in the wireless circuit switched domain so there are different possibilities.
Let's look at standard SIP first that uses the 64 kbit/s uncompressed PCM codec. Add IP overhead and you stream at 80 kbit/s in each direction. Quite a difference to the 12 kbit/s used in the circuit switched wireless network. But wait, it's 28 kbit/s due to error detection and correction. However, that has to be added to the 80 kbit/s as well but how much, that's difficult to say. That depends how far the user is away from the base station, i.e. which modulation and coding is used. So to get realistic values, you have to calculate with a traffic mix. But no matter how you calculate it, there's no way to bring the 80 kbit/s down to the circuit switched value.
Some SIP implementations also use AMR if they detect that both ends support it. That brings down the data rate to 12 kbit/s + IP overhead to a total of 32 kbit/s. For details see this post. Still three times more than 'native' AMR. For users very close to the base station not a lot of redundancy needs to be added so I think we could come pretty close to GSM or be even better. But then, you switch-on half rate AMR and GSM is doing better once again. You could do that in VoIP as well but the IP overhead won't go down and it's already 2/3rds of the total bandwidth for full rate AMR.
Better spectral efficiency could also help to some extent to compensate for higher VoIP data rates as mobiles close to the base station do not only require less error detection and correction bits in the stream but can also use a higher order modulation, thus making the transmission more efficient than GSM circuit switched. But again, that's only for some but not all mobile devices.
Something that works against VoIP efficiency over wireless networks are channel assignments. While circuit switched timeslots are only assigned at the beginning of the call, bandwidth for VoIP calls over HSPA needs to be frequently re-assigned. There were some efforts in 3GPP to reduce the need by using static assignments but it starts getting messy quite quickly here (HS-SCCH-less operation).
But wait, there's IP header compression in UTMS, at least in theory. In practice, however, it's not used as far as I know, so I won't put that into the equation.
Over the top VoIP such as Skype uses pretty bandwidth efficient codecs that are in a similar bandwidth requirement range as AMR. There are lots of VoIP systems that could be used over wireless as well but I don't know what kind of bandwidth needs they have so I won't discuss them here.
There's a real pressure with LTE to switch to VoIP and similar dependencies on features such as modulation and coding, signaling overhead, etc. as in UMTS will have an impact. Robust header compression will probably make it into LTE much faster than in UMTS, be it for IMS, for VOLGA, or for any other network operator voice solution that will be used.
The book from Hari Holma and Antti Toskala on UMTS/HSPA has some interesting calculation on VoIP capacity. Their conclusion is that UMTS packet switched voice capacity can easily exceed that of GSM – if, and that's the big if, all optimizations are present and switched-on. For over the top VOIP, however, it's unlikely that these conditions will be met.
So as you have seen VoIP over UMTS or LTE can be more or less efficient than circuit switched voice over GSM depending on how you look at it. So maybe the question for the future will not be on efficiency but if mobile network operators will in the future continue to be the main provider of wireless voice calls or if over the top voice providers will take a bigger share of the market for which radio network optimizations are not working as efficiently.
Recently, Nokia has announced that they will integrate Skype into the Nokia N97. Reactions, obviously, have been mixed. But I think the trend is difficult to stop, if not on this device it will be on another or in another way entirely. Some network operators have responded by announcing that they are thinking about introducing special tariffs which would include VoIP. But there is one thing over the top VoIP (i.e. non-operator circuit switched voice) doesn't have today, and that is the possibility to ensure the quality of service (i.e. latency, delay and jitter) especially over the air interface.
However, with a bit of imagination it wouldn't be too difficult to set this up. Here's one example of how it could work: In tariffs that take VoIP into account, the network could establish a secondary PDP context (UMTS) or a dedicated bearer (LTE) when it detects IP traffic of VoIP applications. This prioritizes the voice IP packets over other IP packets in the data stream of the user and also over IP packets of other users. Most mobile network operators already have deep packet inspection devices in their networks for all sorts of things and these could easily do the job.
I think it's an interesting technical possibility, let's see if somebody picks it up and puts it into commercial reality.
It seems I am not the only one thinking quite positivity about Voice over LTE via Generic Access Network (VOLGA). Recently, Ajit Jaokar posted an interesting article in which he mentions that with VOLGA, the traditional circuit switched voice service becomes a bearer aware application, as it can choose between a 2G circuit bearer, a 3G circuit bearer and an IP based bearer over LTE. All seamlessly with handovers during the call with all bells and whistles attached!
An interesting way to look at it even more so as the bearer awareness does not come into play on the mobile device but actually in the network. This is because the controlling entity for the voice call, the mobile switching center (MSC), sits in the network and is informed by the network that a different bearer should be selected. It can then decide to go along, arrange for the network to prepare the handover and then instructs the device to make the jump.
So maybe VOLGA makes voice even more than bearer aware!? So far the term 'bearer aware' has mostly been used for applications being aware what kind of networks are available at a time and then make a choice as to which IP network to use or to stay put in case a network is available but the cost attached to it is too high to make the application feasible.
In the case of voice, however, the service can ensure continuity by jumping from one bearer to another. So terms like 'bearer adaptive' or maybe even 'bearer agnostic' come to my mind, because that voice call will just work over any kind of network the device supports.
It could even work over the Wi-Fi you have at home if you extend the idea of VOLGA. Not for the moment, as the standard currently focuses on LTE, but in the future, who knows?
If you have a couple of minutes, join me for a tour of the incredible evolution of the Mobile Switching Center for mobile voice calls from a circuit switched elephant to an IP based leopard. As much as possible, I've tried to use general terms and reduce the number of telecom acronyms to have a blog entry that is understandable to a wider audience. The purpose? To explain at the end why the circuit switched and the packet switched IP world should loose the fear from each other.
Only 10 years ago, the mobile world was a very different place. GPRS was only in the standardization phase and GSM networks were purely circuit switched voice networks. The core component was the Mobile Switching Center, the MSC. A single MSC was composed of several long aisles of cabinets which were full of circuit boards and cabling. Voice calls were circuit switched, 30 of them multiplexed over 2 MBit/s E1/T1 links and many of these links connected the MSC to other MSCs and with the GSM radio network. But if you had seen the earlier generation of voice switches that was definitely state of the art!
Fast forward to today: The MSC still exists and to the user it still behaves like it has 10 years ago. However, today's MSCs have not only completely changed the way they look but also how they work. The switching power of aisles of cabinets is now contained in a single cabinet and the physical circuit switching of 64 kbit/s channels has been virtualized into IP packets running through gigabit optical cables. In effect the MSC has become a powerful server that communicates with the outside world over IP and instead of connecting voice circuits physically in a switching matrix they are now virtually connected by routing IP packets from one IP address to another.
Since 3GPP Release 4, the MSC is logically and in many cases also physically separated into two different boxes with a standardized interface in between them:
The MSC-Server: This is the brain of the voice switching system if you want. When a mobile subscriber initiates a call, the mobile exchanges messages with the MSC-Server which then finds the destination party and signals the incoming call to it. If the destination is available the MSC-Server then establishes a voice connection between the two parties and the conversation can begin once the other side accepts the call. With the old MSC architecture, the Signaling System number 7 has been used for this purpose. Today, the MSC-Server is an IP based system and all messages between the MSC-Server and the radio access network pass an SS7 to IP gateway that exchanges the lower layer of the protocol stack so SS-7 messages can be transported over an IP connection. A mobile network usually contains many MSC-Servers at different physical locations and messages between them are directly transported over IP. Interested in how an MSC-Server looks like? Here's a cool 3D example.
The Media Gateway: The MSC-Server only sends messages through the system via IP links, it is not involved with actually setting up the speech path between two phones. This is the responsibility of the Media Gateway. The MSC-Server is connected to the Media Gateway via an IP connection and the protocol for establishing voice connections is called H.248 or Megaco (Media Gateway Control Protocol). Basically the MSC-Server uses these messages to tell the Media Gatway to connect a virtual or physical circuit coming in on one interface with a virtual or physical circuit on another. I call them virtual and physical because there are different kinds of interfaces. To the GSM radio network, a real physical circuit switched channel is used. Thousands of them are usually multiplexed into an optical link. IP is not used on this link. To the UMTS radio network, it's already a bit more virtual, as ATM (Asynchronuous Transfer Mode) links are used, usually again over an optical link. Also, no IP used here. To other Media Gateways in the network IP links are used and the protocol to transport the voice stream is called Real Time Transfer protocol (RTP). Interested how a Media Gateway looks like? Here's a cool 3D example.
Below I will use the term MSC system to describe the combination of MSC-Server and Media Gateway.
Speaking of the RTP protocol for sending a voice stream through an IP network. It's probably not a coincidence that the same protocol is also used by SIP, which is the protocol used for most of the voice over IP solutions that are in use today. SIP stands for Session Initiation Protocol and if the MSC System is enhanced to understand this protocol, it could also accept and forward voice calls to SIP lines. That's not done in practice today as far as I know but let's keep that in mind for the future.
IP based GSM and UMTS access
Today, the links in most wireless networks to the radio network for the call establishment signaling and the voice calls are not based on IP. However, there are now 3GPP standard enhancements to offer an IP alternative for these interfaces as well for both GSM and UMTS. As a consequence it is likely that over time, network operators will convert these links to IP, too. This could be part of a radio network backhaul convergence, where all data from base stations that support several radio technologies (e.g. GSM, UMTS and LTE) use the same IP based backhaul link instead of several different ones based on different lower layer standards as is mostly still the case today. In short, there is also a trend to replace the current circuit switched based radio network links with IP based connections.
Interconnection between different networks
To allow calls between different networks in the same country and between different countries, MSC-Servers and Media Gateways of different network operators are interconnected. Today, these interconnections are also still based on E-1 circuit switched interfaces and SS-7 signaling. However, as we go forward network operators will start replacing these legacy interfaces with IP links which saves the SS-7 to IP message converters mentioned above reduces cost as E-1 circuit links are much more expensive than using an IP interconnect.
VOLGA (Voice over LTE over Generic Access)
A special case is the LTE radio network. As it is fully IP based system it has no connection to the MSC system today. As reported recently, the VOLGA forum is about to change that as discussed here. Effectively, VOLGA introduces a gateway between the LTE network and the MSC system. To the MSC system, the gateway looks like a GSM BSC or a UMTS RNC. This way, no code has to be changed on the MSC system and the link can either be based on circuit switched or IP technology.
To the LTE network the VOLGA gateway just looks like an IP server and the communication between it and mobile devices flows transparently through the LTE core and radio networks. An already standardized interface is used to announce handovers to the gateway, so the LTE network can inform the VOLGA gateway and the MSC system that an ongoing session is to be handed over to a GSM cell. To the MSC Server, the handover is no different from a GSM/UMTS to GSM handover.
Circuit to Circuit, IP to IP
So when looking at this evolution from a slightly different angle one could say that the MSC looks like to each world what it would like to see. For the circuit switched world, the split MSC-Server and Media Gateway still looks like a circuit switched box. For the IP world, however, the MSC-Server and Media Gateway are just IP based devices which send and receive IP packets. For the voice stream, RTP is used, a protocol well known in the IP world. The call establishment signaling also uses IP packets but the protocol used is originally from the circuit switched world. It doesn't change the principle that it's now IP based but protocols like M3UA, SCCP, TCAP, MAP, etc. are not yet in the toolkit of IP engineers. But fear not, Wireshark and other well known IP tracing tools understand these protocols so they are as easy to trace and analyze as any other IP based protocol. The beauty of the approach is that old and new, circuit and IP can be mixed and matched for each voice call as required.
As we go into the future, more and more links connecting the MSC system will be converted from legacy circuit switched to IP and physical circuits are put into virtual circuits, embedded into RTP IP packets. At some point, some MSCs in the wireless network will only use IP based links between different servers, to the radio access network and even servers in another mobile network. This might still be a couple of years away but the standards are all in place and the changes are well underway.
So from my point of view there is still a lot of life left in the MSC architecture and it will serve us well into the future, possibly even in LTE networks unless a miracle happens and alternatives such as the IMS based system suddenly emerge from hibernation.
Google has been in the press in the past couple of days for resurrecting Grand Central, a web based PBX system that, among other cool features, can forward incoming calls to several virtual and physical phones simultaneously. Some of the reports also say that for the moment Google Voice is only available in the U.S. and not in Europe and other parts of the world, as termination charges to fixed and mobile networks are higher. That’s an interesting twist on the different charging models for calls that are sent from one telephony network to another.
In the U.S., telephony networks use bill and keep (B&K), which means that the originating network can keep whatever it charges the caller and the terminating network routes the call to the destination free of charge for the originator network.
In Europe and elsewhere, calling party pays (CPP) is the standard interconnection charging model. Here, the terminating network charges an interconnection fee to the originating network. For fixed line networks this interconnection fee is around 1-2 cents a minute these days while for mobile networks it’s much more expensive, around 6-10 cents depending on the country.
Some people including me and for example Tomi Ahonen think that CPP is a good idea because:
- Incoming calls on mobile phones are free unlike in the US, where B&K makes the mobile operator charge mobile users for incoming calls.
- This in turn keeps many people from getting a phone, has led to relatively high priced flat rate plans or a high number of included minutes and thus keeps people from communicating more cheaply and getting more than one subscription or mobile device.
- As a result mobile adoption in the US is still far behind that of most other industrialized nations.
- Also, there is no incentive to increase coverage and capacity of networks since the terminating network is not getting paid for incoming calls so it doesn’t matter if they are connected or not. Outgoing calls are included in the flat fee as well so missing or bad coverage does not reduce revenue either.
The Internet Based Voice Centric Service Twist with B&K
So in general, CPP has been quite good for competition and network build outs. But here is a twist with Google Voice and Internet based voice centric services: In the US, while B&K doesn’t generate revenue for Google for incoming calls, it allows them forward calls to mobile networks for free, because the terminator pays for it. In the CPP world (Europe, Asia, etc.) Google would have to pay for those calls. As the users phone number for incoming calls would have to be a fixed line number for the service to be attractive to users, the difference in termination charges they could earn with the fixed line termination fee and the money they would have to spend on the much higher mobile termination fee makes the business model very different.
What this means is that in the US, Google just has to recover its own costs for running the servers and for interconnecting with the telephony network. In the CPP world, Google would in addition have to recover the cost for the difference between the fixed and mobile termination fee, which is anywhere around 4 and 8 cents a minute.
So it looks like this aspect of B&K is quite beneficial for Internet based voice centric service innovation. From a mobile operator point of view, the bottom line does not look as good because when people start using this service their traffic and workload increases while their revenues stays flat due to flatrate pricing. In effect the potential service upsale can easily be made by Internet based companies and mobile operators have no handle to promote their own services, if they had them. Maybe an incentive for them to consider moving to CPP to reduce competition from Internet based voice services? It would be an irony.
For my upcoming course on LTE services at the University of Oxford in April, I've been giving the voice over IP topic some more thought. Unlike UMTS, LTE is a pure IP based network so it doesn't have an inherent circuit switched voice capability. It's a bit like burning the bridges behind you so you can't go back.
From an operator point of view, potential solutions are the IMS (IP Multimedia Subsystem) or reusing the circuit switched MSC architecture over an IP based channel such as promoted by the VOLGA forum, an approach that I think has a great potential! But what are the benefits of network operator VoIP vs. other purely Internet based alternatives such as Skype or about Grand Central / Google Voice, about which Ajit Jaokar has written recently in conjunction with VOLGA? Well, handovers are!
It's handovers because Internet based Voice over IP services will work well while the network can handover a moving user to another LTE or UMTS cell. But as soon as UMTS and LTE coverage runs out and the system is forced to go to the GSM/GPRS/EDGE network that voice call is history.
Not so with network operator based voice systems. For both IMS and VOLGA, methods are in place to prepare a circuit switched channel in the GSM network before the handover and the mobile device is instructed to use this channel after the handover. In the case of VOLGA, it's a pretty straight forward thing to do. Here, higher layers of the protocol stack will not see a difference between the voice call being transported over an IP data stream over LTE and a circuit switched timeslot in the GSM network.
In my opinion, an invaluable advantage for wireless operators that Internet based voice services will not be able to mimick. And I agree with Ajit, in the future we will see application layer based voice services such as Grand Central and network layer based voice services of wireless operators working together instead of fighting with each other.
Over the past two years I've written numerous posts about different proposed options on how to do voice calls over LTE and the lack of a simple and straight voice solution. This is, in my opinion, a serious threat to the success of LTE if not resolved soon. A number of alternative solutions to the IP Multimedia Subsystem (IMS) have been analyzed in 3GPP, which is envisaged to be the successor to the circuit switched MSC voice architecture. However, even after many years of standardization, it has still not seen the light of day and some fear that it's become to complex. Instead, only fallback to GSM or UMTS for an incoming voice call has made it into the 3GPP standards. Looks like some parties in the game are not happy and have started to push things forward by going off stream with the newly formed Volga-Forum (Voice over LTE via Generic Access).
It seems that other solutions, which have been examined in 3GPP for connecting the current MSC (Mobile Switching Center) architecture to LTE have so far not found the necessary approval by the necessary majority of 3GPP members to become a work item, a necessity for becoming an official feature. I've discussed one of the alternatives back in August 2008 here for those who would like to have a closer look. In summary, the two MSC options examined suggest to replace the lower layers of the current voice protocol stack by IP, while leaving the higher layers of the protocol stack untouched. With relatively little work, the existing voice service can be reused for LTE, including the seamless handover of an ongoing call between LTE, UMTS and GSM and international roaming.
Even simple things don't happen over night and each day waiting for 3GPP to finally start working on this topic drags out the day when LTE can really go beyond USB dongles or built in laptop modems and compete with HSPA where voice calls work like a charm. This is where the Volga-Forum comes in. With a strong list of supporters such as T-Mobile and Ericsson, just to name the two biggest (some more operators in the boat would be nice though), they have pledged to take the task of defining an interoperable voice solution for LTE outside of 3GPP.
I would guess the aim is to put some pressure on 3GPP to move forward and the intention is to fold their work back into the standards later on. But even if they stay separate, the solution selected is based on UMA/GAN, as described in option 2 in 3GPP TR 23.879, so the Volga-Forum does not depend on the support of 3GPP to drive their implementation. The main parts of the approach require no modifications in the LTE radio network, the LTE core network or the MSC. All enhancements suggested would be implemented on the mobile device and a new gateway controller between the LTE core network and the MSC. The situation is thus very similar to the days when Kineto and others privately developed UMA, which tunnels GSM and GPRS over Wi-Fi, before it officially became part of the 3GPP standards some years later after being renamed to GAN (Generic Access Network).
A very interesting move, and I think it's good to see that things are moving forward! If you would like to add anything on the political or technical side of this, please leave a comment below.
Thanks to LTE Watch for the first pointer!