 of the details of GSM. So, maybe if you move to slide 29 somewhere around there is where we have been talking about that there would be an MF mobile node, BTS is basically what we understand as the base station, BSE is the base station controller, MSC is the mobile station controller and so on. So, we will go a little bit more into details about how the GSM air interface is first structured and then we will come back to look at how roaming works and other such issues. Again, let us not go through all the slides sequentially. Let us just go ahead and try to do it as though it did not exist and we are inventing it. So, just to quickly recap we have now GSM operates in the 900 megahertz band. So, we know that which is basically 890 to 915 megahertz and this is the uplink 935 to 960 megahertz downlink. We have seen that it has 125 channels, 200 kilohertz. This is what is called the physical structure of the system. This is the physical structure of your GSM system. Now, we want to understand how this what is the logical structure of the system. Let us try to ask a few questions. One is what happens during power on, outgoing call, incoming call. See, the reason why we are trying to do this in this manner is that as it is, it is very dry. There is lot of acronyms and lot of really dry description and if we take it at 2.30 in the afternoon after a nice lunch, then I can kind of anticipate the effect. So, let us try to do it in a different way. Let us try to understand why it is set up like this. So, that is the objective. So, what it is can be found in a book. Why it is something which is a little bit more involved. So, again keep in mind the objective. What is the objective now in a GSM system? I am paying a lot of money for licensing. Always keep that in mind. I am paying a lot of money for licensing. I need to extract the maximum juice out of my system in order to make profit and so on. So, now what we are saying is. So, think of it this way. Suppose the power on everything is established and I want to establish an outgoing call. Would I allocate one slot straight away? So, the question is like this. If you see this is how the, this is what we have understood so far. 935 megahertz to 960 megahertz is the downlink which has 125 channels. Each channel is split into 8 slots. Similarly, there is a 890 to 915 megahertz. Each channel is split into 8 slots. So, let me ask a question here. Why is 890 to 915 the downlink sorry the uplink and not the downlink? So, the reason here. So, the point I want to illustrate is that all of these numbers have some logic behind them. So, it could well have been the other way. I mean if we are just doing it in theory it does not matter which one is uplink which one is downlink. But it turns out that what you want to do is you want to conserve the battery power of the mobile for as long as possible right. So, now if it is transmitting at a lower frequency it is going to consume less power. So, that is the reason why you have the uplink as 890 to 915 megahertz and the downlink is at the higher frequency because that would require more power, but the base station anyway is powered up. So, that does not matter as far as the base station is concerned alright. So, given that we are doing like this. So, the question now is that as soon as somebody wants to establish a call do we just pick up one of these slots and allocate it to that mobile alright. See the key reason is unless you are sure that the call is going to happen there is no point in tying up the entire resources. So, you want to delay tying up of resources for as long as possible right that is the key idea you want to delay this tying up of resources. So, how do I decide whether the call will be established or not? What are the parts involved in establishing a call? This mobile has to request for a channel then the network has to see whether the where the other guy is right and then the network has to figure out whether that it has to give a ringing tone on the other's other side that guy has to pick up only then the conversation happens. So, there are several steps in a call to take place correct and there is no point in allocating a traffic channel what is called a traffic channel till all these steps are successful correct. So, each time I just simply request something if a slot is allocated then that slot goes could go waste for several frames ok. So, that is the key idea here that logical channels are split into what are called control channels and traffic channels. What is a logical channel now? What is a physical channel? One of these things is a physical channel time slot 1 in frequency number 123 on the up down link. So, that is how you will define a physical channel. What is a logical channel? The logical channel is the same physical channel which repeats every once in a while right it occurs here then after 8 slots it is going to occur again. So, when you put all those things together that is what you are kind of calling as a logical channel ok. So, if I say that frequency number x time slot y is your traffic channel you know that every time when frequency x comes in time slot y I am going to transmit my voice packet ok. So, that is the meaning of a logical channel alright. So, one thing which we have discovered is in order to extract the maximum utilization out of our system we need to split our channels into traffic channels and control channels right. So, now the question remains is to figure out how do we utilize these control channels ok what are the different control channels that we need ok. So, these three questions will actually help us to figure that out. So, what happens when the mobile node powers up ok what are the things that I need to know upon powering up power on. So, what do I need to know I need to register with the register with the network ok. How do I do this? How do I know that the network exists when I power on beacon heartbeat. So, basically what will happen is the first thing is base station has to transmit something like a beacon ok. Now, this beacon occurs in almost every type of wireless technology in one form or another ok. Sometimes it is called a beacon, sometimes it is called a hello packet ok, sometimes it is just called a broadcast packet alright. So, base station has to transmit some something like a beacon ok. In this case it is not a beacon in GSM system it is not called a beacon it is called a broadcast control channel ok. So, how do I know that I am receiving the beacon? Is the question making sense? See finally, all said and done it is a string of zeros and ones. How do I know that I am receiving a beacon now? There are so many channels. How do I know which channel the beacon is being transmitted on? I am the mobile node ok. There is a base station out there. I know that there are 125 channels on the downlink. I know that the mobile node the base station is going to be transmitting the beacon on one of these channels. How do I know which one? So, either I could have a fixed channel assigned for transmitting the beacon correct. So, base station has to transmit a beacon. So, the question that arises. So, if you try to answer these questions see what is happening. So, for each question we are coming up with some answer and then some other question associated with it that is how the whole technology develops ok. So, base station has to transmit a beacon moment you say that we ask the question which channel ok. So, this is just one step of the registration. In that itself we are already asking the question which channel. Now, for which channel we have two options right. We have fixed frequency x times lot y ok. Suppose we decide that frequency x times lot y will be the channel on which I transmit the beacon ok. What is the implication of such a decision? One has to wait till wait till what time? Time slot is just 577 microseconds right 1.4 microseconds 1.4 seconds may it is going to come I have to scan the frequencies do I have to scan the frequencies. If I know that this is frequency x I just go to frequency x right. So, those are not drawbacks. So, what is the drawback? There is a dedicated channel. So, what is the plus points of this is that this is easy to find right. It is easy for the mobile node to find that frequency x channel y ok. But if you see what are the minus points of doing such a thing ok what is the key minus point? One channel will be wasted because I have to keep sending this broadcast information all the time ok that cannot be helped. What could be a key drawback? See we have 125 channels for the entire spectrum ok. How many operators are there in any given area GSM operators are all 125 channels made available to every operator right. If I make all 125 channels available to every operator what will happen? Two of them will put up a base station in the same area right and then you are going to have interference right. So, that is not the way the allocation happens. So, the allocation will happen like this the government will say ok I am going to give these many channels to one operator, these many channels to another operator, these many channels to a third operator and so on right. So, that is the that again is done by the telecom regulatory authority ok. So, for example, in European countries you have about 18 channels per operator, 18 channels are given to each operator ok. In India it is about 9, each operator gets about 9 channels in which to manage the system right because all the operators together can have to stay within this range. Sometimes the if the number of operators exceeds the number of channels then what do we do ok then that is a bad regulatory mechanism ok. Then you will keep on increasing the licensing fee right till some of those guys find it economically unviable see that is the game right. You keep on making it harder and harder making the entry level harder and harder. So, that only a few guys can finally remain see it is a very limited resource that you are playing with right. So, that is the game you have to play ok. At some point you are going to finish operator however successful may find that is run out of channels what does he do that is why you have frequency reuse. That is why you make smaller and smaller cells you put up more and more base stations all that fails sometimes you might say that is the maximum capacity of the system right. So, coming back to this. So, now can you tell me what is a key drawback of fixing this frequency x times lot y. So, every operator will have a separate frequency that happens even today, but if I hard code frequency x times lot y into the mobile then I cannot do I cannot do roaming you see what I am saying. If I hard coded it into the mobile that frequency x times lot y then I can only work with that operator only in that area I cannot roam because the other operator is going to use a different frequency A and times lot B which I do not know about right because this is power on right. If I could receive the beacon from the beacon I could say ok this is the frequency on which the broadcast is happening, but before receiving the beacon itself I do not know. So, this is like you get stuck at the first point itself. So, this is no you cannot do any roaming if you are going to use a fixed mechanism of identifying this is that making sense to everybody ok. So, we agree that we should go for a dynamic mechanism right yes or no yes. So, given that how do I find out which channel is the beacon coming on same problem. So, now I know that I need some dynamic mechanism I cannot do fixed how do I find out which channel is the beacon coming on I scan the channels and no code what I cannot find any format. See think of it like this you have 8 slots So, which means I have to scan each channel scan at least 8 slots in each frequency channel then move to the next channel then move to the next channel then move to the next channel that is the way I have to move in order to find out whether there exists a beacon in my system or not is there a easier way of doing that. We can fix the time slot within every channel not really done, but it is not I do not think it is incorrect to do that. So, what is done is very straight forward see again you want a very simple solution what is that what could it be. So, channel on which there is the highest power see because that is what the mobile has in its hand at that moment it cannot decode any signals it does not understood it has not synchronized with the base station right. It does not know how far away it is from the base station it does not know what is the clock frequency of the base station nothing, but it can always measure the received signal strength because that is a very very basic measurement which can be performed by the device independent of whether it is registered or whether it is connected or anything right. It has to scan through all the 125 channels or whatever all the channels of that operator okay if it is that is why whenever you power on it takes a while for your you know that antenna stick to appear on your mobile right. The signal strength stick takes a while to appear every time you power on because the mobile has to scan all the channels and then it looks for the channel which has the highest signal strength highest received power that too it does not know that this is the broadcast control channel. It assumes that this may be the broadcast control channel and then tries to decode the information if it is able to then decode then it knows that okay I got it right otherwise it goes to the next highest power channel and next highest power channel and so on okay is that making sense okay. So, let us say we figured out this thing so this is called the broadcast control channel BCCH okay this is the most important control channel in your GSM system okay what is the information that has to come on this channel okay. So, the information that comes on this channel is frequency correction okay lots of these acronyms exist okay there is frequency correction first you have to align in frequency with the base station right you may be slightly off I mean these are all very low level electrical stuff okay you may be slightly off but it will make a difference. So, you align in frequency and you have to synchronize with the base station okay synchronization channel. So, there is a frequency correction channel there is a synchronization channel what other information comes along with the BCCH what is the information suppose think of it as a packet that is being transmitted think of the beacon as a packet which is being transmitted by my base station what will go inside that packet what should go inside that packet cell ID okay then timing reference location means what cell ID subscriber is the mobile node the complement of the subscriber right network ID operator ID okay. So, this is some of the key information which your broadcast control channel has to carry okay what is the cell ID what is the operator ID so it has to say orange this that what not otherwise the how does the mobile know which network is it listening to correct. So, cell ID operator ID then it will give a frequency correction and then there is a timing reference channel okay what happens after that we will come to the details of how the timing happens okay what happens after that so the base station has transmitted the beacon then what do I do I have synchronized in frequency I have synchronized in time okay synchronizing in time is not very easy we will get to that so that I know where I am is just a BTS ID cell ID or the BTS ID it is called the BTS ID also okay. So, the next step that should happen is so assuming that the timing has been synchronized so the step 2 next thing that it does is after the so basically what has to happen is mobile okay so the first thing that happened in power on was that the base station told the mobile that it exists correct does the base station know that the mobile exists no so that is the next thing that has to happen in your power on scenario the mobile has to know somehow notify the base station that it exists how always means which channel at this point right which channel do I send this information the mobile is just powered on it scanning all the channels one by one it finds one channel which has the highest power it says that okay this may be the broadcast control channel it monitors that channel to see that is there a frequency correction sequence that is coming in this channel once that comes it says okay I am corrected I am synchronized in frequency with the base station then it looks for some timing information so that it now knows what is the need for the timing information so that it can it can align itself in time with the base station right so what the base station thinks as slot 3 should be the same as what the mobile thinks as slot 3 if they are different then there is going to be chaos in the system right somebody is going to transmit at some point somebody else is going to you know transmit at some other point right so all that it has managed to do so it is now aligned in frequency it is now aligned in time now it has to tell the base station that it exists reverse control channel correct we need some reverse control channels so how do we set up these reverse control channels that is the question right so how is we need some uplink control channels or reverse or uplink okay some reverse or uplink control channel that is not its name very different name randomly I try to transmit in any channel you know all you people are speaking I just come there I power on at randomly transmit what happens to your conversation I cannot do that you know the question is not making sense I guess so I need to find out which channel in which to transmit my uplink information how do I know which channel fixed again we have figured out right fixed does not work fixed will lead us into some problem or the other as far as the even for the beacon we tried the fixed as the simple thing correct so channel information will come along the BCCH so that is all along with operator ID I will say which is the uplink info channel okay it is not called the uplink info channel I am just calling it like that for ease of understanding okay okay let us say so let us say that frequency X time slot Y was the BCCH and I go into the BCCH and I find that for the uplink info it is some other channel you know frequency A time slot B okay so I know that this is some channel how do I transmit on that channel suppose two of us have powered on at the same time what kind of mechanism should I use random okay so because the very straightforward thing we do not really care if two people collide then let them back off let them come back and try again so basically this is what is called your random access channel okay or the RACH this is the key uplink channel okay so the random access channel uses slotted aloha as the mechanism okay slotted because we are in a slotted system aloha basically means you just randomly get up and transmit that is all it means right aloha means you just transmit slotted aloha means you transmit at the beginning of the slot you cannot transmit anywhere in between that is all right so in the random access channel I am going to send my registration packet correct so what are you going to say you are going to say that I exist that is all you can say because you do not want to tie up that channel for very long either right you do not want to stay on the random access channel and prevent somebody else from transmitting two things can happen you might or somebody else may transmit and corrupt your data so you can send only a small amount whenever you are sending information at random in a broadcast medium you can send only small amount of information so basically all you can say is that you exist and you ask for a so it is something like you say I exist and then you ask for a another control channel okay information exchange channel let me call that as information exchange channel right you say that there is no I do not want to do a lengthy information exchange with the base station on a random access channel so on the random access channel I only notify the base station that I exist now give me some channel on which I can talk to you okay that is all it means give me some other channel on which I can talk to you okay so what is the third thing that happens okay this is called the dedicated control channel okay so you are saying that give me a dedicated control channel on which I can exchange authentication information with you I can exchange you know privileges capabilities all those things away from the random access channel okay that means sense okay so what is the next thing that should happen channel assignment so this channel has to be assigned right so BS assigns dedicated control channel correct BS can assign the dedicated control channel how just picks one you know n number of channels available to it just picks one and says okay this is the dedicated control channel so that is the easy question the harder one is how does it inform the mobile load right assigns the dedicated control channel and notifies correct so it can assign it by just picking one which it has 10 channels which are free at that point it may just pick one of them and say okay this is the channel that is free and it can assign it correct so how does it notify it back to the mobile load now that we know that there are so many different types of channels you just say okay there will be one more channel in which it is doing that correct that is basically what happens so it notifies it on what is called the access grant channel so the access grant channel has to come in the beacon so the beacon has to contain all this information so now you know when you look at a beacon you know why each of those items are there okay the access grant channel information has to come in the beacon so that once the mobile node has sent its request for the dedicated control channel it's going to move to the access grant channel and keep listening you know is there any access assignment which has been allocated to me okay then what happens then it's pretty much straightforward see once the access grant channel is there then both of these guys know that dedicated control channel is whatever frequency m times slot n okay so everybody knows not everybody the base station and the mobile node know that they are going to exchange their control information on this frequency m times slot n so they go ahead to that they exchange the control information whatever hlr update has to happen that happens right whatever authentication has to happen that happens right it checks whether it has roaming privileges checks whether this is authenticated base station right you may even have fake base stations right even the mobile node has to authenticate the base station so all the authentication everything happens and that is the stable state after the power on okay which is the dedicated channel in the downlink okay what is the significance of such a question see the significance is that this channel is in both directions so far we had the broadcast control channel is only in the downlink direction we have the random access channel only in the uplink direction right again access grant channel only in the downlink direction right but the dedicated control channel has to be in both the directions because I need to send information to the base station base station has to send information to me right and now what I have written here is that dedicated control channel is frequency m times slot n okay how do I know which is it how do I say which is whether this is for uplink or for downlink so one way of doing it is very straightforward see the brute force method is always straightforward right I could always send two different information to the mobile node I could always say that this is the one for uplink this is the one for downlink that's easy but that's tedious can we do something smarter that's the question how do I know which one whether to use the same one so if you see this figure you see that there is something which is shown as a delay okay so what is done is for any in a GSM system any two way channel we use the same frequency number and the time slot number okay see for example this is frequency channel 123 okay this is also frequency channel 123 okay so the channel ID is the same 123 but the frequency in which they are operating is different right right this guy is in some 914.8 to something this guy is from 959.8 to something but they are both frequency channels so this is frequency channel 123 on the downlink this is frequency channel 123 on the uplink okay so when I am assigned frequency channel 123 times slot 3 as my dedicated control channel or as my traffic channel it also means that frequency channel 123 times slot 3 is my corresponding uplink channel okay so that coupling is understood that is the that's where they use a little bit of intelligence so now what happens here is now both of these cannot be happening at the same time right I cannot be receiving and transmitting at the same time so there is an offset between the uplink sequencing and the downlink sequencing so there is an offset of 3 so 3 slots so what this means is if I am receiving data on frequency channel x times slot y then on the corresponding uplink frequency after 3 more time slots I will transmit the data so I don't even have to keep track of which is the slot ID I just know that okay I have received it in this slot after 3 slots I am going to transmit okay that's the way it works so again it's a simplification for easily understanding or easily computing which is the channel okay all right let's see what happens now when we have a incoming call which one shall we do first incoming or outgoing incoming okay okay let's do an incoming call incoming call means let's say from a PSTN somebody is dialing your mobile number okay how does your mobile ring first thing is base station sends it on the it has to page for the mobile right what we just worked out before going for lunch I had to do some paging in those cells saying where is this guy where is mobile x okay then allocation of dedicated channel for response how do I know which cell to allocate in I am paging in multiple cells paging will always happen in multiple cells see what is happening the mobile the dedicated control channel is also not going to last forever right the dedicated control channel is also given for only a certain amount of time till the authentication everything is done then you want to release the channel why do you want to release the channel so that that resource is available for traffic correct you don't want to tie it up so moment the mobile hears paging on random access channels right so the mobile will say will try to access the network on your random access channel again this is kind of a I exist message okay it's basically going to send a message on the random access channel that okay here I am right then access grand channel okay then dedicated control channel okay okay now is a good time to introduce its full name sdcch so sdcch stands for stand alone dedicated control channel okay so on the access grand channel it's going to say let's talk sdcch okay that's basically what the base station is going to say let's talk on sdcch and then sbs okay sdcch is a two way channel in which you will again you know exchange capabilities and then we'll do ringing okay we'll say that there is an incoming call okay so the mobile will ring and then when this guy pick up okay so when you pick up that's when the traffic channel is established okay the traffic channel gets established once you say okay pick up the mse is going to page two different bsc's the bsc is going to also page in all its cell okay but when it comes to the cell it's the base station which has to transmit the request on a particular frequency and a particular channel so that is called the paging channel okay is that making sense bcch right so all that information has to come in the bcch which is the paging channel which is the access grand channel everything it has to know so once it has finished authentication which channel is the mobile monitoring paging channel right once I have finished authentication I finished everything then I'll go and sit on the paging channel in case there is an incoming message in case of an outgoing call that doesn't matter because the device somebody is going to dial on the device and then outgoing can start from this point correct mobile will just send a random access channel message so this is so the outgoing is actually a subset of the incoming as far as the air interface actions are concerned right is that making sense suppose it's outgoing mobile is going to send a message on the random access channel saying I exist give me a sdcch okay that's what it's going to say on the random access channel so then the base station is going to give it an sdcch on the access grand channel on the sdcch what is going to happen is now it's going to say dial some xyz number okay some ringing is going to happen at the other end okay which may be a psdn number which may be whatever number basically that information it's going to just pass on to the base station right then this is like ringing at other party and then finally you're again going to come back once the other party picks up you're going to get assigned a traffic channel okay tch is basically the traffic channel tcs is voice data exchange before that there is no tch for transferring voice data that's when we are going to use the tch okay now so now just see if you understand everything now okay base station there is one base station per cell so the base station provides basically signaling channels data channels okay there is a random access channel okay base station is identified by some base station code okay base station controller does all these jobs of radio resource management this is what we saw earlier okay time and frequency synchronization signals are sent to the base station time delay measurements are made okay power management is done all right mobile switching center is going to do all these mobility of subscribers and all these kind of things okay I think we have finished with most of these topics gateway connects to the mobile network to a fixed network okay coming to the air interface we have seen all these things 124 channels of 200 kilohertz everything okay okay there is a notion of a burst burst basically means packet okay so what we understand in ethernet as a frame what we understand in ip as a packet okay what we understand in tcps a segment is called a burst in gsm so there are four different types of packets normal burst is the one which corresponds to data transfer okay so look at this so the normal burst looks like this it has a 3 bit header so how does a typical frame look in ethernet I have a header I have a payload and I have a I have a crc yeah I have a trailer so why does this frame look different so see what what you have we have a 3 bit header 57 data bits there is one bit here there is some 26 bits of training sequences another 1 bit here 57 data bits another 3 bits and some 8.25 guard bits okay why does it look like this why doesn't it look like the ethernet itself you know a header followed by data followed by a error correction okay see so let's say you have a signal or a sequence or a bit sequence okay and you want to transmit this over a wireless channel okay so typically what happens is in the channel you will have some noise that will get added to your signal correct some noise will get added to the signal in the channel right so how do you know what is the channel condition the way to know it is to pre define a sequence right so I will say in the beginning itself I am going to transmit some let's say 1 0 1 1 0 1 1 0 1 1 0 1 1 some sequence like that okay and so this is the transmitter this sequence is what I am going to transmit some noise is got added and what does the receiver receive receiver may receive let's say you know 0 1 0 1 0 1 0 1 something like that it receives right now the receiver knows that it was supposed to receive this okay this is called the training sequence okay so the receiver knows that it was supposed to receive this training sequence but instead of receiving the training sequence this is the received sequence right so if I compare both these sequences I can estimate the noise in the channel right if it is the same then I have a good channel right there is no noise in my channel if there are one or two bits off I know that there is some noise in the channel so I can estimate the channel condition so appropriately I may have certain error protection or error recovery mechanisms okay is that mean sense okay so that is basically what we are trying to do with the training sequence here okay so the training sequence is for channel synchronization or estimating the channel condition okay everybody agrees why does it come in the middle typically you would put it in the beginning okay because it is a wireless channel on the other hand if I put it in the middle I can assume that more or less this is what held here and more or less that is what holds here so by and large I can recover both parts of my data if I put the training sequence in the middle instead of at the beginning okay that is the key reason if I put it in the beginning then it will hold for certain amount of bits after that but then it may drift off if I put it in the middle I know that it holds for what came before and holds for what came after that is the reason okay is that make sense okay so what are these one one bits here what does it say says one signaling bit so those are actually not separators they are flags which are called stealing flag okay the stealing flag is like for emergencies when you want to send control information in a data channel okay when would you want to when would you have an emergency as far as a mobile is concerned when there is an imminent handoff right suddenly there is an ongoing call and suddenly I find that the base station with with whom I am communicating the received power has gone down drastically so there is an imminent handoff but I cannot wait for sending a packet on the associated control channel because that may take a while to come so I just have to send it immediately in the data channel itself so at that point what the mobile does is it sets one of these flags so once the flag is set what it means is the following data is not voice data but it is control data okay so that is what is the meaning of a stealing flag okay okay what is 8.25 guard bits 8.25 is not guard bits actually it is guard time okay so since we are saying 577 microseconds and if you convert all of this it will come to slightly short of 9 bits it is 8.25 so it is not like do not think of it as bits but it is a certain amount of time which is left vacant on the channel guard time it is okay okay so on the air interface we have seen that there is a traffic channel broadcast channel control channel dedicated control channel there is also something called an associated control channel which is basically used when there is a traffic channel existing if there is no traffic channel then I live only with a dedicated control channel if there is a traffic channel then there is an associated control channel okay so I guess most of these things you will know okay so let us come to one interesting aspect okay so ignore all these things full rate traffic channel half rate traffic channel and all those things so what we want to understand is let us go back to traffic channel okay okay traffic channel we saw that there are 57 bits plus 57 bits okay in each frame we have 57 bits of data and then there is some other 57 bits of data that coming okay so what we will try to understand now is what is it that we need to keep in mind why is there no CRC see when do I need CRC I need CRC for error detection right so there is no use for me to do error detection again here see there is a trade-off now because my channel is a very important resource I do not mind putting in extra effort to do error protection I do I cannot afford to have error detection in my system because error detection is always followed by typically followed by retransmission okay so I do not want any retransmission in my system that is a waste of channel resources okay so in a GSM system there is error protection instead of detection and retransmission okay so the idea why it is to be done is fairly straightforward I want to save my I want to just simply avoid retransmission I do not want to do any retransmission but at the same time I want maximum amount of the data to go across right so now the question is how does it do error protection okay so one thing we have to understand is that it does not transmit the voice as it is okay so what we what we may assume is that have you wondered why your friend's voice the same friend's voice sounds different on a landline versus a mobile versus real why is that so how many of you have noticed that is true right many people have noticed right why does the voice sound different in each system okay so why what happens with voice voice I am going to sample okay so I am going to sample voice and I am going to generate a bunch of bits okay so if it is an analog system or if it is a wired system it does not matter to me I send the whole thing I send the full sample let me just say speech is sampled and in a PSTN line this can be packetized or in any other voice over IP line it can be packetized or digitized or whatever right and transmitted that is a typical way in which I could do that all right what happens in GSM again I do not want to waste my bits on the air so I do not want to send all the sampled bits okay so in GSM what it does is after this sampling okay it has something called a model chooses what are called vocoders okay voice coders okay so this has a model of human speech voice box okay so what instead of the entire sampling being transmitted what it is transmitting is basically vocal cord frequency X excitation energy Y okay just transmitting key parameters okay so these are the parameters that are being transmitted so instead of transmitting the entire sampled speech it just models this voice and it transmits saying that okay for this frequency of vocal cord frequency so much is the energy that you have to supply and so on so these are the things that go and at the other end okay so at the other end the speech gets synthesized again you have the same model it just supplies that for that particular vocal cord frequency it supplies that much excitation energy and that is why your friend's voice sounds different because this is synthetic speech it is not the sampled speech as it is which is what you would have heard in this case correct okay so what we now have is a bunch of bits now okay what are our bunch of bits we have bits saying corresponding to the vocal cord frequency and the different excitation energies okay so now see what happens in your GSM system okay so what it does is this is how it does the error protection first it does speech coding for 20 millisecond segments which results in 260 bits at the output okay after I do all this vocoder business 20 milliseconds of speech after sampling and after going through the vocoder comes to 260 bits at the output okay now out of these also certain frequencies are more important right certain other frequencies are less important of the human voice right now so what it does is it does what are called unequal error protection what is the simplest error protection mechanism repeat right so that is the most simplest error protection mechanism you transmit the same information twice so you find that in this case out of these 260 bits you have 182 bits which are protected okay so basically 50 bits are taken and you add 132 bits of protection information to it to ensure because these are the critical 50 bits to ensure that these bits are not lost under any circumstances if those bits are lost then you cannot hear at the other end you cannot make out what you are hearing the other 78 bits which are unprotected they are only going to add to the quality of the voice that you are hearing but they are not essential so you leave them unprotected okay so then what happens is the channel it cores the 260 bits into 8 into 57 bit blocks of 456 bits okay let me come to this in a moment okay so see what is happening you have speech 20 milliseconds of speech is taken to 260 bits okay 260 bits is encoded into 456 bits is that making sense channel encoding is where you are going to you can say kind of add the additional modulation information everything okay so this is basically what you have to transmit on the channel 456 bits have to be transmitted on the channel okay how many levels of error protection are over already one level of error protection is over here because in this 260 bits we know that 50 bits are very important and they have been you know protected by adding them twice and so on okay so now we come to this 456 bits what does it do with the 456 bits now that is also very interesting it does something called block interleaving so I have 456 bits which come from the first 20 milliseconds of speech I have another 456 bits which come from the second 20 milliseconds of speech okay so now this 456 bits you chop into 8 bits of 57 bits okay 8 8 chunks of 57 bits alright so this is this blue thing is the first 57 bits from here this blue thing is the first 57 bits from here okay so you remember that in my normal frame what did I have have two 57 bits in the packet that I am going to transmit right so what it does is it takes this 57 bits puts it in the first half here of the normal burst it takes the other 57 bits and puts it in the second half of the normal burst understand so in one packet what it is doing is it is not even sending two speech portions which have any connection with each other it is sending 20 milliseconds from what I said 20 milliseconds ago and it is sending 20 milliseconds and it is sending 57 bits from what I am saying now okay why again if one of this is lost because human voice in the human mind can reconstruct the voice even if one of these 57 bits is lost that entire 20 milliseconds is not lost is that being sense so these 260 bits from here one 157 bits have come here from here 157 bits have come here that is what is going in one frame okay so this is called block interleaving okay so it has done one level of error protection okay so error protection the first is 260 bits I have 182 bits protected correct second one is block interleaving okay which is which basically means 57 bits of two different speech segments right third one is again another interesting thing which it does which is called bit interleaving okay so the third thing is actually straight forward so what it does is with these 57 bits okay what does it do okay it writes the 57 bits into an array in a row wise manner okay and read column wise okay what is the effect of this what happens as a result of this is that see all our speech error recovery techniques work if there are one or two bits here and there which are in error okay if a huge chunk of bits are lost which are contiguous then the error recovery mechanism becomes hard okay so I am basically going to write my data in this manner okay so these are actually our contiguous bits okay those are the contiguous bits right so when I read it column wise what happens contiguous bits get separated out okay now when this goes out onto the network even if there is a burst error here what will I do at the other end at the receiver's end this is the transmitter end this is the receiver end right so in the receiver end I am again going to write it column wise and read it row wise correct so because I am doing this what happens is any errors that are burst errors in the channel show up as random errors in my block okay so these are the three important ways in which jsm tries to protect against errors okay one is doing error protection for some of the data itself okay by doubly transmitting it or whatever then it does block interleaving and finally it does bit interleaving at the before transmitting on the channel okay so that is how we get here okay so since most people are desperate for a break I will stop here