LTE: APN, PCO & Initial Attach

This post is out of an interesting conversation I had with one of my readers.

In LTE, UE gets attached to a network in a single shot. That is when the UE is switched on it will send a NAS message, Attach Request along with PDN connectivity request. Look here.

3GPP TS 24.301 has a little different story to tell though. According to the spec UE shall not include APN and PCO in the PDN connectivity request when the same is sent along with attach request. The spec says UE shall send the PDN connectivity request with a flag "ESM Information transfer" on and no APN or PCO shall be included. Once MME receives the Attach Request+PDN connectivity request, it shall go ahead and accept the attach but it shall not establish the EPS bearers just yet.

MME now goes ahead with establishing security context. Look here. After the security context is established MME will send a NAS message "ESM Information Request" asking UE for APN and PCO. Now UE shall send "ESM Information Response" with APN and PCO, encrypted(?). Once MME receives this response it will go ahed with establishing the EPS bearers. If the response doesn't include APN then default APN shall be used by MME.

Why this? I guess this is for security. We just dont want to reveal the user name passwords to any network that asks for it, right?. And also above is only valid if Attach Request and PDN connectivity request messages are sent together. The story is different if the two messages are sent separately. More thoughts?

LTE S1 Handover: Indirect tunnel

I wrote about LTE S1 handover in the white paper. However I dint cover this interesting concept. I am working on next revision of white paper and will make it public soon.

X2 based handover is used when there is X2 link between source and target eNB's. This also means that EPC should be just informed about change in eNB FTEID's for downlink. If we look deep, the downlink packets are still sent to source eNB until EPC receives a handover complete notification and the target eNB FTEID's. The buffered downlink packets are sent to target eNB by source eNB later which are then sent to UE. These packets are directly sent over the X2 link. This means there is direct tunnel formed between source and target eNB for data forwarding.

Now I started looking at a S1 based handover. I saw something called Indirect Tunnel.

Now in S1 handover we assume that there is no X2 link between source and target eNB. Also I am assuming that SGW has changed during the handover. So the downlink packets buffered at the source eNB during handover execution should be sent all the way to Source SGW, then to target SGW and then to target eNB. To do this target MME sends a GTP based message Create Indirect Data Forwarding Tunnel Request to target SGW. In this message the target eNB FTEID for downlink are sent. In the same way the message is sent from source MME to source SGW. This will create a indirect tunnel between source and target eNB. After this the buffered packets are sent from source eNB to target eNB which are later sent to UE.

Sounds good? Anything to add? Please feel free.

An observation!

There are bunch of office boys at the place I work. Since one year I have noticed them using big fat mobile phones while I carry a small GSM phone which can do nothing but voice and sms. So I talk to the boys and they tell me that these phones are made in China and apparently are very cheap. They also told me that these phones might stop working anytime or they might just keep working for years.

Later I observed them using GPRS and accessing WAP sites of service providers. They download a ringtone or a photo of an actress everyday. Now this made me wonder. People who have never used internet on a computer, or rather who have never used computer, have started using internet on mobile phone. So what happens when these guys realize the beauty of youtube. If 3G comes in and service providers start advertising youtube service on television, these guys are going to have a blast. Remember its the adds on television that made facebook so popular in India (Aircell advertisement?). This means India is directly moving to mobile internet, which is jumping over a step, skipping the internet revolution on PC. India is adding 10 million subscribers every month. Now with 3G coming and youtube like services rolling out just imagine how congested the networks are going to get!

Any thoughts?

IMS is the way to go?

I have been strong supporter of VoLGA, but I have recently realized that industry should move towards one long term solution. I believe IMS is the way to go in future.

3GPP clarifies the LTE Myths - Strong support for IMS :-)

LTE Dedicated Bearers: The Big Question!

I wrote about LTE dedicated bearers some time back(here , here and here ). Just when I thought I got every thing right I realized that I am back to square 1.

We know that dedicated bearers are network initiated. But the big question is when does the network decide to initiate the dedicated bearer. Many people have asked me about this so I thought of putting down what I knew.

In LTE there is a bearer which is always established until UE is shut down and it is called default bearer. Well UE can ask for a dedicated bearer by sending out a bearer allocation request to the network. Once network receives the UE request then we will have dedicated bearer. This is one case.

What we noticed is there is a trigger for dedicated bearer. Now if we look at the spec for dedicated bearer creation, it immediately starts with Create Bearer Request. Many are confused with what could be the trigger for this message. Next para should help.

Second case is:- default bearer is established. Network now wants to have all the http traffic put in separate dedicated bearer. So once the default bearer is established, network might ask UE to run the http traffic on dedicated bearer. Note here that there is no trigger for dedicated bearer, which means network is configured in such a way that all http shall run on dedicated bearers with a particular qos. Even if UE is running http of default bearer it may be asked to switch to dedicated bearer. If UE is not capable enough to run a http then it may reject the dedicated bearer request also it may reject the request if it thinks default bearer is fine for http.

Now the confusion here is IMS calls. Many say how will network initiate a dedicated bearer for IMS calls. The big confusion is with reception of IMS call. Well, this is what I think. IMS needs an application to be running UE, a special app. If UE is intelligent enough then it will request a dedicated bearer for IMS call while it is registering with network or when the IMS app on UE starts running. Else PGW may look at the packets that are being sent by UE (deep packet inspection?) and initiate a dedicated bearer by it self. The point here I want to make is we shouldnt think that network will create a dedicated bearer only when UE receives an IMS call, the bearer is created much ahead of it.

If we consider that network is modifying or creating a new bearer for already running traffic, then what about the UE application's connection state. Will the connection be reset as the bearer has to be switched. I dont know, an UE engineer should answer this. How ever this dedicated bearer is still pretty confusing, in fact many are confused with what could be triggers for a dedicated bearer creation. Any ideas?

Evaluating LTE SGW

We know that Serving Gateway is one of the most important entities in LTE network. Say if XYZ service provider is evaluating a SGW, what are the factors that XYZ would be looking at. Lets avoid price and technical support offered by the equipment vendor here.  

Serving Gateway is supposed to handle both control plane and user plane traffic. Just to increase the complexity, lets consider that S5 interface is PMIP based. So we have GTPv2 communication on S11 interface, GTP-U encapsulated traffic on S1-U interface, PMIP based control plane and GRE encapsulated user plane traffic on S5 interface. So the technical requirements are revolving around control and user plane.

Control Plane: Input to the SGW are GTPv2 messages and output are PMIP message towards PGW. In between SGW needs to communicate with PCRF/Radius servers for pulling other information. So first thing XYZ wants to see is a successful session established. That means we need to get the basic functionality correct. More over XYZ would also want to see how easy it is to configure the SGW. This may not be one of the criteria for evaluating the box but its good have a easy to use interface. Getting the basic functionality right means there needs to be a successful interop. That is the SGW should work with other vendors MME or eNB or PGW or etc. This also means that SGW vendor should get the specs right :-).

Next, XYZ wants to see how many sessions can be activated per second. Also XYZ would be very interested to see how many active sessions can the SGW handle. May be a million users per box? That means SGW should have million contexts activated and data base integrity should be smooth. XYZ might also want to see how many dedicated bearers can be activated for single default bearer. More over XYZ might fancy to see how many default/dedicated bearers can be activated and deactivated per second. This test might prove how strong is the box is.

Handovers is another functionality which XYZ wants to see. Does SGW support various handovers mentioned in the spec. If so XYZ would prefer a number of users that can have successful handover per second.

Roaming too is important. There is typical requirement which arise with roaming scenarios. MME of visiting network might want to contact home network SGW for authentication, charging policy etc. Which means the GTP message is hopping for one router to other in the internet. Which also means that there needs to be some security involved. IPSec comes into picture. MME can encrypt the GTPv2 control messages in IPSec and pass it on to the home network SGW. This means SGW should also be to handle several IPSec sessions. I dont want to go below this layer. SGW might also be able to do BGP/OSPF/MPLS etc at the network layer. These could be some of the control plane requirements.

User Plane: First thing that comes into my mind is throughput. XYZ wants to see how may user plane session can be handled at what throughput. Usually the packet size is fixed in S1_U interface, so there is need to know the maximum throughput that box can achieve. I strongly believe that this is the most critical factors.

Finally, XYZ wants combination tests, thousands of sessions established along with thousands of user plane tunnels with different kinds of traffic. These are system level tests. If everything matches XYZ's requirement then they are ready to spend heck loads of money on the gateway.

These were few thoughts running through my mind. Feel free to add more!

IP CAN Session Establishment

BBERF :- Bearer Binding and Event Reporting Function PCEF :- Policy Charging and Enforcement Function H-PCRF :- Home- Policy Charging and Rule Function V-PCRF :- Visited - Policy Charging and Rule Function

Do we need LTE any sooner?

Ok, I have been thinking real hard on this. When do we exactly need LTE? Next year? Two years from now or three years from now? I am trying to understand various article which try to justify the right time to launch LTE, but I could read none of the articles completely. Either they are too long or they go over my head. So I thought I should do it in a simple way.

Lets break things in to two here. First, what do we baldy need that is lacking now. Second, what do we fancy/love to have.

Note that first need is something that is lacking us from doing things while second one is more like wow to have or something like that.

Before we categorize the needs we need to look at the market segment. That is how many people need that badly and how many are fancying it. The things which few people badly need could be fancy to few.

If we replace "we" with "I" in above statements it would make more sense to various individuals as the next generation is all about personalization.

So if you ask me I dont need LTE. Honestly I dont see its need for me atleast in next 5 years. I work in office where we run pretty good internet connection. I get back home and have a nice wifi to use. I really dont have big fat mobile so I skip my emails if I am traveling and can live with it. I am not a facebook addict, infact I dont know why I am still using it. I prefer to call up people and have conversations rather than facebook messages or what ever. SMS is the best way for me to send a message. In case I need to be hooked on to email all the time I would go get a black berry and use exiting 2G or 3G network. The only thing I fancy is a data card. It would be good to carry a data card hooked on to my laptop so that I can be at a remote place and still be connected. This is the case with me and also with tons of my friends. What good is LTE is going to do to me? In fact the whole corporate breed in India can fall into my category.

I want to hear from you. Do you need LTE? I would really appreciate your comments on this. Cheers!!

P.S: Please dont ask me to go back to stone age. I understand that few people do need mission critical applications, that need high data rates, running on their mobile devices all the time. I am just trying to figure out what is the right time to deliver LTE considering the huge costs involved with it. I would get a heart ache if LTE fails because it was launched too soon or too late.

I will add three more constraints. Available spectrum, back haul and number of devices. How will the time lines look considering these too.

LTE Tidbits

Few more interesting things on LTE.

GTP & PMIPv6

So, looking at the market I came to know that PMIPv6 is the one that will be deployed over S5 interface, no matter wether the existing service provider network is GSM based or CDMA based. Even I have been PMIP supporter for a while as it avoids complexity over S5. Reduces complexity? I was thinking over bearers one day and realized something.

In LTE a UE can have 11 bearers altogether established per APN. Note that EPS bearers are GTP based bearers. Now on S5 interface there is no concept of default bearer or dedicated bearer. That is there is only one bearer (PMIP) per UE per APN. The question is how are multiple bearers on S1_U interface mapped to single bearer on S5 interface. Lets call each bearer as a pipe with different quality of service running different applications. Since each pipe is identified by a TEID we can enforce all the QoS on it over S1_U. But how will the same be mapped over S5 interface? Interesting? Any clues?

Hint: There is an entity called PCRF which is responsible for bearer establishments, QoS enforcements etc. So SGW contacts the PCRF or rather PCRF informs to SGW on how to handle each GTP pipe over S5 interface.

More later.

Handovers & Tracking Area Update

Have you noticed that there are no NAS messages sent for handovers? That means all the handover decisions are taken by eNB based on the power measurements etc and UE is informed to modify its RRC connections.

But if you take a look at 23.401, there is a call flow for Tracking Area Update which gives us an illusion of handover. This is special case I guess so it is separately dealt, not as handover, in the spec. The difference between handover and Tracking Area Update is the later is UE initiated. Each MME(or SGW?) has a list of tracking areas which it tracks. This list is sent to UE during default bearer establishment. If UE detects that it has entered a new tracking area that is not present in the list sent by MME, then it will trigger a Tracking Area Update procedure.

Radio Bearers and EPS Bearers

If you look at the default bearer and dedicated bearer establishment procedure there is an interesting fact hidden with respect to radio and eps bearers. For default bearer, EPS bearer is established first and then the corresponding radio bearer is established. But for dedicated bearer radio bearer is established first and then the EPS bearer. Before I prove it, this fact leads to another two interesting facts. They are MME is responsible for assigning EPS Bearer Identities and we need modify bearer request for default bearer establishment. Let prove all the three.

Looking at default bearer establishment, Create session request is sent from MME to SGW after it receives Initial UE message with PDN connectivity request+Attach request. Create session request includes the EBI and same is informed to SGW. SGW sends the response if the bearer is accepted along with its user plane information. With Create Session Response the EPS bearer is established (EBI is assigned & SGW user plane info is known). Later MME goes ahead with establishment of Radio bearers for the same using the EBI. Once the radio bearer is established (eNB user plane info is known) the same is indicated to SGW in modify bearer request. This proves the third fact.

Next, take a look at dedicated bearer establishment. Dedicated bearer is network initiated. That means Create Bearer request is coming from SGW and it contains LBI, SGW user plane info, TFT etc but no EBI (set to 0?). Once this message is received by MME, MME assigns an EBI and goes ahead with establishment of radio bearer. Once the radio bearer is established (eNB fteid is known), then MME informs the same to SGW in create bearer response along with the EBI. This proves another fact that MME has to assign EBI, but not SGW.

So the above two explanations prove the first fact.

Why EBI is of 4 bits?

EBI is 4 bits because NSAPI is of 4 bits. NSAPI is used to uniquely identify a PDP context in GSM/UMTS networks. To maintain the compatibility EBI is also set to 4 bits. This means there cannot be more than 16 values for EBI. Out of these 16 values 5 are reserved which explains why there cannot be more than 11 bearers. Dont ask me my NSAPI is of 4 bits :-)

Thats it folks, I have run out the ideas. More to follow. As usual any corrections or comments are are greatly welcome.

LTE Backhaul

Let me start by wishing you all a very happy diwali.

Long back I wrote about MPLS being considered as backhaul for LTE. I believe it is still the choice. More random thoughts on the same below.

Many have said that with LTE bandwidths will explode, infact I say even with HSPA the bandwidths will explode. Lets not look at this from GPRS or UMTS or LTE perspective. The simple logic is if there are many users even with GPRS, bandwidth will explode if the link between SGSN and GGSN is thin. What is happening now is there is only one IP link in wireless core network, so there is not much emphasis on it. The case is same with HSPA. With LTE there are several IP links, devices that run on IP have increased and more over relation between these devices is many to many. The wireless core architecture is moving close towards the wired architecture.

Many IP devices bring in a Central management unit along with various OAM tools. So if Verizon wants to deploy LTE, how is it going to choose the EPC vendors. I dont think using MME from from NSN and SGW from ALU will be a good idea. This also means that EPC solution providers should also come up with Central management units for their devices. One place to configure and monitor the devices.

Next challenge is to keep up the service level agreements and end user quality of experience. This is very very challenging. I will touch base on this pretty soon.

Another challenge is the actual backhaul it self. While the fiber will be widely used there are other backhaul techniques like microwave and copper. I am not sure how many microwave links we will see, but yes, there are some solutions based on these links deployed and are running fine.

Next point is infrastructure sharing. Atleast in India I have seen BSNL lending its network to private operators. The same can be done with LTE.

Quite a few things lined up, it would interesting to see how they will shape up.