UTRAN Protocol Stack... Quick help needed!!!

[hot_ice]

utran protocol stack

Hi!
Well, I need to know the Uu (UMTS radio interface) interface protocol satck. I've studied it, but I'm a little confused. I know, it works at Physical, Data Link and Network layers of OSI model.
I want to know, what exactly are the responsibilities of these layers. I am going to write what I've understood so far. If I'm wrong please correct me.

Physical Layer: This layer is implemented in Node B. And mainly, it is responsible for transmission at radio. It is controlled by RRC (layer 3/Network). Any other responsibilities?

Data Link Layer: This layer is subdivided in 4 layers: MAC, RLC, BMC and PDCP. MAC is responsible for managing multiple access to a given resource.
RLC is ensures correct data transmission.
BMC sublayer is responsible for point-to-multipoint communication.
PDCP sublayer is responsibe for IP data transmission. It performs header compression.

Network Layer/RRC: This layer is solely used for controlling the other layeararrs. How???

This is all, I've concluded so far. But, I would like to know more about these things. If any of the layers/sublayers have some other responsibilities, please do highlight them. And, one thing more, I need quick reply. Anticipating an assertive response.

Added after 3 hours 5 minutes:

And, please try to explain the differnce between physical, logical and transport channels. Thx!

 

[gied]

utran stack

you've got it almost correctly.

Physical layer (or layer 1) main responsibility is data transfer over radio (air). And that's a lot. Of course it has lots of many other functions, (fast power control, spreading and data coding, interleaving, etc.), but those functions are essential for data transfer over air. It depends on the viewpoint - you could say that they are part of data transfer.
Moreover it would be more correct to say that Physical layer terminates at NodeB. The physical layer is implemented in UE as well, just as the all the other protocols you have mentioned. However the remaining protocols terminate at RNC.

Data Link Layer: just few statements, that could give you better insight
MAC - it makes decisions on who/what and when should transmit. Who/what are allowed to access system.

RLC - it is mostly for packet data - acknowledged transmission together with restoration of correct order. Voice or video will RLC transparent mode - it means RLC is switched off.

BMC - DVB-H will use this.

PDCP - it only perform IP header compression. that's it- nothing more.

Network layer:RRC - this is network control protocol. It uses messages to tell UE what to do. For example network will tell UE what it should measure and report. When measurement is done, UE will report results with another message. Then network would tell UE to establish another radio link with different NodeB. And that command will come in RRC message.

Considering different types of channels:
physical - the one that are transferred over radio.

logical - these are differentiated on what they carry. Traffic channels carry user data, control - network control info.

Transport channel are defined on how the data is transferred. There are common and dedicated channels

Both logical and transport channels are subchannels of physical. For example BCH is common transport channel that carries system information, however it is not sent separately over the air - BCH is combined
with some other transport channels and sent over P-CCPCH (a physical channel).

Logical and transport channels have been invented just for easier reference on where and how network data is transfered.

I hope i haven't confused you more.

 

[hot_ice]

node-b stack

ty gied!

I need to know some more things... Physical layer is implemented at Node B, no doubt about that. Where are the other layers implemented... My logic says, the other layers must be implemented at RNC. Is that true?

 

[gied]

mac-hs header example

Yes, all the rest are implemented in RNC. By rest I mean the one you have mentioned. There are PPP, CM (call managment), MM(mobility managment) protocols that terminate in Core network, i.e GSN or MSC.

There is one exception to that rule. However if you consider HSDPA/HSUPA, then you will see that there is MAC-hs, that is implemented in NodeB. MAC-hs schedules data transmission in HSPA.
Scheduling decisions should be done really fast in HSPA - RNC could keep up with such a rate. So, part of MAC has been moved to NodeB.

 

[hot_ice]

rrc protocol stack

Thx again!

I'm about to ask another silly question. Believe me, this one is really a silly Q. Anyways, here I go.

Why do we need QPSK or 8PSK for modulation, when we already have deployed CDMA technique.

As far as I think, by multiplying user data with the chip, we get another chip. Now the new chip has to be transmitted, but we cannot transmit it as it is in pulse format (i.e. its a digital signal). To transmit a digital signal over the radio interface we would need infinite BW, which, we obviously cannot have. So, in order to transmit, we use QPSK. Is that the actual story???

I do not hesitate in asking questions, but, on the other hand, I admit my concepts are quite weak. So, please help me out. Thx!!!

 

[gied]

nodeb protocol

The pleasure is mine.

First it should be stated that CDMA is not modulation but the method of multiplexing channels or data from different users. There plenty of other multiplexing or multiple access methods - tdma or ofdma just to name few.

For data transfer over RF modulation is necessary. It is possible to do without modulation, if transmission media is a wire or a bus. It is necessary to confine data into certain frequency bandwidth - it is done by modulation.

Both user data multiplexing and modulation are necessary.

When you are considering user data with the chips, you are spreading data to wider bandwidth and multiplexing later on. The resulting bandwidth is from 0 hz up to 0.5 chip rate. (it is simplification, valid only for binary spreading. It is unsuitable for transfer over air. WCDMA uses complex spreading - single chip has 4 values, rather than 2 as binary case. Furthermore depending on PN function it is possible to have different bandwidth that 0.5 chiprate.

Now modulation shifts data to desired carrier frequency and bandwidth.

I have little knowledge why Phase shift keying has been chosen over frequency or amplitude. I assume that PSK is more suitable for direct sequence spread spectrum.

There is very little difference in SNR performance between BPSK and QPSK, but QPSK allows double data rate. 8PSK and all the higher rate modulation schemes have worse SNR performance. Usually they are not used.

Another point to be noted is that QPSK is just the same as 4QAM. Next higher order modulation is 16QAM, that has higher data rate than 8PSK but similar SNR performance. This is the reason why it has been selected for HSDPA. The problem with 16QAM is high requirements on PA linearity. It is hard to make small powerful PA with high linearity. HSDPA employs 16QAM only in downlink, as NodeB can be much larger than UE.

You are absolutely right that digital signal has infinite BW. The same is true for QPSK, however infinite bandwith happens only at phase jumps, that is when one symbol(chip) changes into next. To limit BW, pulse filtering (filtering in time domain) is applied. WCDMA uses root-rised cosine - lookup in the internet or wikipedia for further information. The idea behind pulse filtering is to decrease amplitude at chip transition instant. Effectually that decreases out of band transmission levels - it could be said that it makes narrower BW.

I suppose that this explanation is not too difficult or too easy for you.

 

[hot_ice]

umts layer 2 utran protocol stack

Well, in my view, deploying WCDMA means we are multiplying the user data with a chip (which generally is different for each user, and enables us to spread the user data on wider BW), and thus, it features multiple access/multiplexing. The new chip obtained is in pulse format.

QPSK, the modulation scheme used in UMTS, is used for pulse shaping (if you look at its mathematical expression, you would come to know that the modulated data is no longer in pulse form, rather its in analog form -- And, hence,we can transmit it over radio interface).

As far as, Root-Raised Cosine filter is concerned, it merley is used to avoid ISI. It also is available at the Rx. side.

QPSK is better than PSK, as with the given resources (BW), data rate for QPSK is better than PSK. And, also, you are right in stating that 8PSK and higer modulation schemes are not used. We use QAM for that matter.

 

[oldandfool]

gied, these are the best explanations I've ever read about UTRAN protocols. I had been very confused about them until now, but I have a clearer view now. Thanks.

 

[nitinpuri]

I have one doubt. I agree with all the talks but i am not sure from where does the Iu-b stack of protocols come into picture with Radio Network layer and Transport Network layer. We know that Node B works on

---------- Post added at 15:11 ---------- Previous post was at 15:11 ----------

I have one doubt. I agree with all the talks but i am not sure from where does the Iu-b stack of protocols come into picture with Radio Network layer and Transport Network layer. We know that Node B works on WORKS ON PHYSICAL LAYER ONLY. SO WHY O MUCH OF STACK?