 So, you are actually looking at the time space time configuration. So, I will just go through the control structure which will be implemented for this. So, this actually now technically we are going to talk about control of distal city and I am talking about certain a specific variant which is T S T configuration. So, I am going to now look into only T S and T configuration and of course, I am going to assume that it is right cyclic, read a cyclic on this side, write a cyclic, read cyclic. So, that all control memories you will be accessing the same location at any point of time. So, usually what you will be controlling in this way that is the first question. So, that will actually decide what kind of actions have to be applied. So, it is like kind of building up what we call basic instruction set of a machine. So, there is a basic instruction set of a switch also correspondingly. So, basic operations you have to identify and what basically we are doing is there are control memories for all three systems and what we are doing is we are just putting in writing into those control memory. Rest everything keep on happening and this is being done by the circuits all the time. So, that is our objective and there basically three components which will be there. So, these components are the first one is switch itself you take a time switch is control memory how you write into that how it is being read out that is the control interface. So, whatever is the address the bus data bus address bus everything which goes into control memory that is your interface and that is a switch control which we have already talked about while looking for the implementation. Similarly, for a space switch again it is a memory it is in a column it will figure out which particular junction have to be switched on in which particular time slot. So, again it is a control memory that is switch control again third stage time similar kind of structure. Next we are going to always have a switch block control. Now, this whole combination itself can be called as a switching block a larger size switching block and we would like to communicate there will be master command which will be now sending information to individual switch controls leading to the whole switch block to be controlled. Each individual has to be separately referred or separately the information has to be pumped into that and then of course, that is a switch block control usually will be implemented in hardware, but then switch block control has to be given information by somebody who is looking into destination address how the call has to be set up and then based on that it will be doing some kind of routing calculation to whom it has to be forwarded everything. So, that most intelligent software entity running in the system has to talk to switch block control. So, that particular control is known as exchange control system. Of course, with CIP telephony all this has been now done away with we actually have technically removed switching and transport functionality separately it has been done totally separately. Currently transport is much more tightly integrated with the switching as of now, because I am setting up a transport path there you do not do any setup of transport paths switching runs in IP layer it is independent of transport. So, far you can set up a TCP connection between two entities you create a transport, but you do not tell IP layer what has to be done what not has to be done here both are tightly integrated because it circuits system. There you actually create an overlaid circuit over a packet switching system that is the only difference. So, that actually creates more decoupling there. So, these are three things and now the switch block control which is this I am not bothered about this I will talk about the message types I am more interested in this we have already done anyway. So, there actually now categories which you will be doing through this you have to somehow set up the path from incoming port to outgoing port. So, first function will be always set up of a path I am now giving a sample which is a hypothetical sample as of now, but it is very close to reality in real switch implementation whenever a manufacturer does it he will make his own listing. So, if you are actually going to design a complete switch from scratch you have to build up your own listing of what all operations will be required. May be when you will build up your first version it will go into use you will figure out something is missing some more functionalities have to be added you will even add those and redesign the switch version 2 will come. So, iteratively once it becomes stable you release it as a product actually. So, this one of those identification, but it is more or less what almost everybody does and the forward and reverse paths are set up separately. So, that your switch is much more general usually you will say voice circuit is always going to be bi-directional if I talk to you also talk to me, but then that is not a general switch. Tomorrow I might require I might have a requirement where I talk and 100 people will listen, but the problem is remaining 100 people when they will talk back there will be so much of even the noise which is being picked up. So, those 100 people I will some of the noise itself is going to be higher than signal actually. So, usually reverse channels have switched off at that point of time or they will be routed through a different place and then it will be selected one of those depending on who is active depending on the voice activity I require some such that kind of structure. So, I should be able to set up each direction separately that usually will the way I will be designing because it gives me much more flexibility in design. When I want only point to point voice circuit to be set up I can set up front and reverse simultaneously, but as two separate connections. So, setting of path so, unidirectional and bi-directional there are two options in this case. So, there are two kinds of messages. So, this message has to be given by exchange control to the switch block control. So, there is exchange master command which is going to so, there is switch block control which controls these and the exchange control is sitting here. This is the message which has to go unidirectional as bi-directional both options have to be there. Control is switch control that is the switch control last. But control memory will be written by switch block control. Yes. So, depending on control memory switch anyway is getting controlled by the circuit by the hardware. Second thing which you will be doing is of course, once you set up you have to clear down. It can also be unidirectional or bi-directional both. And I think we should expect very similar things also should be happening in SIP based system, but SIP is I think more general it is not only voice, but many other media types also which can be hooked on to. Sometimes for example, a call is not through still you have not started billing voice has not happened, but you would like to reserve a path for certain time. You have not set up the complete path it is not being, but you keep it blocked because if you do not block it and you send some the next guy you set up the signal you set up the signal you set up the signal. And by the time the confirmation comes this path has been already taken over by somebody else. Most of the real life switches are not strictly known blocking they are all blocking switches. So, that is why if you decide that I would like to try to set up this particular connection you have to make a reservation then send the request forward. The next exchange will also again do the set up and forward the request. And once the confirmation comes in the reverse direction all reservations gets confirmed actually. And then you do the set up of path. So, path gets cleared and every they will start actually working the status bit will actually be changed for every port or every path actually. So, you require reservation of paths again this can be unidirectional or bidirectional. This guy is informing you. Is there any restriction on being making a reservation or anybody can do? Everybody will be making a reservation for that? There is no everybody there is only once exchange control which is commanding the switch block. See I am not talking about you telling somebody to make the reservation. As a result of your interview you are talking to some line card here the information is going to be pumped on to this exchange control. So, whatever signalling you are doing your phone is doing with the exchange is never being given to switch block control directly no. It is being trapped or even PRI there is a signalling channel that is being trapped and goes to the master software. This in turn is going to send the message to switch block control. There is no direct thing you cannot make direct things. You talk with your own language which is specified for you. This is user to network interface and UNI specifications. That is a generic term being used depending on kind of network this UNI specifications will be different. For you you have now DTMF tone. You look for a dial tone you will dial number that is only limited thing which you can do nothing much. He is doing he will figure out whether I should actually make a reservation or not. If the path itself is not available I will not make reservation. Suppose if this particular port is busy I want to send it to some say Delhi. Delhi all lines are poor ports are busy or anyway there is I anyway will not be able to forward the request. So, I will deny the call there itself. So, I will not make any reservation. It knows from the status table that these ports are busy or it can inquire about the status of the call various ports which are toward Delhi. This reservation is within the exchange or between two exchanges. Within the exchange I have to set up the path from input port to output port. I have to make a reservation before I forward my request to further set up. See technically what you are doing you are not only reserving these, but you are also reserving this line outgoing line also to the next exchange because this cannot then cannot be used this slot cannot be used by anybody else to route the call to next exchange. When you make a call how the call is made question is this. The call is your instrument talks to the box the exchange exchange if it sets up a path send a request here sets up the path the question is I want to actually make sure that I can clear it pretty fast. This is one of the ways is you set up the path you keep on and then inform the next guy he sets up the path he informs the next guy he sets up the path and everything. Usually this is way it is never done. Reservation means those tables nobody can write they are logged currently it is like same afford. I will write the final value when things are ok, but nobody else is permitted to write because I am still trying to set up because this guy talks to him this guy talks to him this guy talks to him this does and then he picks up the handset that time the confirmation comes and this confirmation back propagates and this will remove the ringing tone which you are listening at that point of time. And all reserved path will be then confirmed and the path will be set up immediately. See problem is the path you have to identify you have to understand what is happening when the path is being set up from input to output port. You have to identify which particular intermediate slot is available you have to identify lot of things you have to do computation and this computation does take time once the computation has been done. Now, let us take two scenarios first scenario is let there be no reservation something comes to me and I am not bothered I know this is ok. So, I say this particular line I can use. So, I send ok I probably can set up because input output is free and some resources as of now are available I have not blocked them. I just estimated the resources by their status it is available you set up the connection you then it keeps on happening till it goes to the last guy. Nobody has set up the connection they have only identified resources that it is possible to set up there is no confirmation it actually puts a ringing current this guy lifts the handset once the confirmation comes now it is job is to set up the path. It only knows about the IO ports which have been free not the internal paths. Now, everybody has to do a computation once the computation is done this confirmation will be happening in a path will be set up again a computation is done path will be set up computation and path is set up once the path is set up then I will remove the ringing current the delay will be higher while if before even I apply the ringing current I make the reservations and there is another possibility when you come back in the reverse direction the resource itself might get consumed by some other call there might be contention because you have not locked the resources. The same resource is going to be used by same medallist switches can be used by for multiple circuits for setting up of them. So, when you try attempting it is like railway reservation booking there is only two tickets available you say I want to book you do everything you make the payment your ticket is not confirmed by the time unless you make the payment and confirmation happens by the time somebody also tries it and he books the ticket first you would not get the ticket by the end of payment you will be very furious. So, in fact what happens is when you are doing this railway reservation booking you actually now reserve the resource immediately before the final confirmation the move to start the booking process at certain point of time when the status there it remains locked for certain period and by the time nothing goes no reservation happens when you do not come out it will be released automatically for the next booking. So, there is a some period where the locking actually happens it is similar kind of thing locking of resources in the reverse direction because resources are already locked and identified you have to just flip the status bit path will be immediately start functioning you keep on it happens very fast and you can immediately remove the ringing tone which is coming back and two people can start talking and of course billing will be generated by this guy always billing record hunting is a part of that is why the hunting always takes more time, but once the guy other guy lifts the headset it does not take time. This will be part of reservation. Yes, but how many lines in exchange percentage wise reserve certain amount of lines one line for each call which is being made if a call is not complete within certain time the resources will be released automatically. So, those status are again the control memory which I have shown I have not actually shown other fields their extra fields attached to every word memory word which is there which I even talks about the status because remember location in the control memory specifies path it actually tells a middle stage path it tells you exactly what is the status. So, there actually some other extra bits which are there which can be used in the design, but again there is no standard on this a standard is not required because this whole box will be manufactured by one single manufacturer. You are not going to buy control memory from somebody else and switch from somebody else and put that things together you buy this whole thing from one single guy you buy another thing from another person these two should be able to talk. So, in standards are only when two different vendors are involved interfacing between them single vendor thing you do not try to do anything because you leave it to the you know the imagination of the vendor or his design team will keep on innovating inside you cannot be too rigid actually that is the way the industry actually has evolved. So, reservation is essentially in the forward direction is. So, most of your hunting delay which happens is because of this. So, it means all the calls get through they get through by reservation of parts only. First the reservation happens and reverse then there is a confirmation when the setup actually happens that is the process where there is no other possible way I think. In fact, this does not require any technology this is all common sense best ways you say I am going to design my own system. In fact, if you have ever have any confusion you do not have books you think I am going to design the system I how I would have designed the system and 99 percent chances you would have designed the system exactly in same way the way it is already being implemented. So, lot of actually you do not need to look into manuals to understand the existing system because of this. So, these are common sense while in the reverse path setup there will be delay. So, these guys will get annoyed actually always even the reverse confirmation I start putting up the delay. So, forward setup you keep the delay there is no issue, but the moment you lift the handset it has to be almost instantaneous because you have to start also doing the billing at the same time. You will be losing otherwise these precious seconds for which you could have built you are actually increasing the load. You can do the peak computation and increase state before the coming the confirmation from the reverse direction. Usually there is even a better and a smarter thing is done is not cascading thing what usually is done is the moment this guy talks I have not actually talked about this because this signaling is if it is a in band signaling in band means in the voice circuit itself you do the signal we call it channel associated signal. If I am using a separate packet switch network which for signaling which is what the SS7 does this guy once it figures out a call has to be set up it estimates a route it can send a packet parallel to each one of them there is a parallel transmission which happens all of them will make the reservation of resources when the packets will come it will be for certain duration and they will respond back with packets actually to this guy if it figures out I have got all the responses the resources are through on all the whole path is now clear it just simply will send a message to him to put up the dial tone sorry the ringing current the phone will ring here once the ringing confirmation happens it picks up it comes this will simply do is nothing will send the parallel packets to all the elements parallel again through SS7 SS7 is like TCPIP similar stack 5 layer stack they do not call it a physical layer network layer data link they call it MTP level 1 MTP level 2 MTP level 3 and so on. So, only the name changes of the layer functionality wise they are all exactly same they also use same kind of routing principles whatever we did in 673 and then they have already made the reservation is happening parallely now the call setup is still faster now that is a difference between your 20 years back or 25 years back if you would have made a call and now you make a call there is a difference and difference is because of this most of our signaling is now on SS7 it is not CAS earlier days it was actually still worse you might have waited for talking only for 1 minute but signaling and call setup might have taken 5 minutes actually. So, I do not know these older systems I think are no more operational anywhere except probably in some of your locations where you do not want to put in the money but in India I think almost we are fully on SS7 and now I think in next 10 years even SS7 will be done away with fully and we will have purely said in between H.323 would have been there but now it is all discarded out H.323 is not picking up because there is open source battery equivalent which is there with version 3 running currently as of now SIPV3 if you have and of course with IPv6 and SIPV3 if the both combinations is I think is a good option always because you have QS support available in IPv6 which is required for this is called media transport actually all voice transport is a media transport it is not data transport. So, anything which is real time is a media transport whether it is a voice video does not matter and for the first time all kind of media can be transported by same signal which is actually not permitted. For example, you cannot make video calls with the conventional telephone it is not possible because that support was never it was basically only for voice but SIPV is general you can actually extend all kind of media types. Now, next the D part tracing of path digital circuit says does not make sense except you try to find out whether control memory that memory location has gone defective or not there might be a byte which might go bad in a control memory that only means that switch is actually still functional only now the number of possibilities or intermediate options are going to be less. So, you should not use those combinations. So, this basically tracing of all possible combination this basically used for diagnostics. So, again unidirectional and bidirectional both E is this is the performance basically checking of path we call it but basically performance analysis whether something has gone noisy or not typically on the boards. So, you will be actually putting some source at the input and something at output you set it up you will check the performance. Tracing is you will read you will write and then read back kind of thing to whether things have been written correctly or not in the control memories. These were exactly also setting up the path you are looking even at the wiring timing synchronization everything will get verified. What time do you do the tracing of the path? Whenever the load will be less that time really it will be done you never do all the diagnostics and everything are never done all the time no usually whenever the you will actually find out in the day is a daily kind of variation time variation when the load is least that time you will put the resources for all diagnostic test. It is done usually automatically because this can be done by this guy this guy sending a message corresponding to these events. What happens if we discover something? He will just generate a log report he will at least give alarm somebody can come and then repair it otherwise how you will know you cannot go go and manually keep on checking all paths. See if you are going to there are exchanges of the size of the like 1 lakh lines almost in some big cities 50,000 is pretty common 50,000, 60,000 all exchanges were built of that even size but they were having multiple switch blocks even IIT Kanpur here we are going to have actually now currently I think 20,000 lines that is what we have requested 7,000 will be commissioned by next month and currently operating exchanges 6000 line extendable to 10,000 which is now going to be actually taken out apart from the service actually. The numbers are used 6000 line manually is going to be impossible in fact our fault rate should not be very large because I cannot sustain man power it is not commercially viable to have large man power to maintain. Number of phone lines at the homes which goes bad has to be very, very small. So, number of phone equipments which goes bad has to be extremely small. So, typically the harsh conditions are put we buy a complete telephone equipments the condition is that within first 2 years if more than 2 percent of the phones actually goes bad then the whole lot has to be replaced by the manufacturer under that condition only you buy the telephone equipments. So, you have to give a warranty for that we do not repair we do not have that much of time and telephone equipments is something which is currently is being maintained usually by the provider but I think now scenario is going to change usually people should buy company only provides you a port at your home that is it. But I think still in India everywhere it is the operator which provides you the phone instrument but they usually outsource it to directly to manufacturers and he will be making loss. So, I think the production quality of the equipment has to be extremely good if failure rate is too high it will become non viable because ordinary equipment on ordinary telephone not ship based. For us if we buy and bulk cost about I think 175 200 200 rupees roughly in between that that is when we purchase with all honesty another government define the cost must be slightly higher. But here because I do the negotiations I know the price. So, next is F this is interrogation. So, whatever is the status of any path which has been set up whether it is free busy clear reserve. So, there are various status for every path. So, they have to be modified. So, modification is done through these set up cleared reservation these three things basically are used for setting up all the status. But these status words also have to be queried back when you are for example, trying to set up a new path you have to periodically now synchronize your local status table with whatever is there actually inside the switch. So, these status are also maintained in the control memories here. So, this interrogation is typically will be for the free busy reserve three things three status. Part of form of the interrogation and reservation will remain together. Now, this you are actually now setting up the values here you are inquiring about the values you are inquiring what is the current status. I mean the first interrogation then reservation or. Yeah usually that is the way interrogation is usually you are most of these the way it will be operating it maintains a local table. Usually it will be it will not be inconsistent because all set ups and all release are done by this. But the time there is always a possibility I have to work with that I cannot be unidirectional. I have a database before I always send a message I will always make the changes in the entries here. Chances are that most of the time whatever is my status says is actually what is going to happen in reality at the other end. But the other end also can go wrong actually some once in a while which you will find out for checking and tracing all that thing I can still make my local thing. But I think I will as an engineer I will keep an extra precaution I will always run a separate thread in the software you will keep on querying the actual status and see if there is a mismatch if there is a mismatch corrected running a small thread is always makes sense I think you all understand what is a thread. So, thread is a separate what we call sequence of execution which runs in a program a computer actually can run multiple threads same EXC same binary can run multiple separate threads which run independently of each other. So, it is like multiple CPUs running but technically one CPU it is time shared all the threads run separately. So, same thing also we have to run it is like computer. So, exchange the way I always say is nothing but a highly specialized computer which does all the time computation but of a different kind. These are also known as stored program control switches SPC that is another term common use I have never told that. So, you have now six categories. So, what will happen is there will be six of these there will be total twelve kind of messages this technically means because is a bidirectional unidirectional combination is there. So, six into two total twelve possibilities and there these message type need to be identified. Now, these messages are between this and this exchange control to switch block control switch block control to switch I have not done so far. So, usually there will be three kind of type one messages will be going in this direction but remember this is manufacturing specific. You need not follow exactly this thing you may design your own because whole thing will be still bought built by one single vendor. So, does not matter whether you use type one type two how it is type one is which is going from exchange one to exchange control to the switch block control type two there basically message formats will be different type two is this and type three is done given actually over hardware lines on a distributed bus on a bus structure. So, type three is this. So, there is no actually no message being sent over a serial communication line this is a parallel lines this is over a serial line this is actually like a packet structure proper packet structure. So, left side bit is always sent first and the lower order bit is always sent later on kind of thing the way IP packets or other packets are identified. This one is a parallel hardware structure. So, I just quickly go through the generic message types here. So, this is the type one message this always goes from exchange control to switch block control you have to and remember this direction this is important the reverse direction is type two and whatever you have studied in 673 or digital communication networks also applies here similar concept whatever we have done there you will actually probably appreciate some of the things because fundamental set will never change wherever you apply that you will always remain same. So, you will typically will have the message number every packet has a sequence number because if I transmit a duplicate I need to identify that and when I respond back I have to tell I am responding to this particular message number and what typically the message will be the opcode of operation code of one of these twelve options. So, next is opcode operation code is very similar to how you build up the instruction set. So, I have technically making instruction set are getting. How is it twelve options? Unidirectional and bidirectional two options are there in each one I think they are total six values total twelve six into two. So, next is opcode opcode will be representing one of these twelve options. You want to set up certain thing you will write I want to do a set up and it is unidirectional or a bidirectional if it is unidirectional source and destination makes a matter and if it is bidirectional source to destination and destination to source both have to be done. Usually there will be some it is not that software is not running here it will be like a firmware which is sitting in it is not on a computer. It is an internal microprocessor which is sitting inside on the board which is going to run the show with limited capabilities will not be exhaustive computing system. So, you are always when you you are inserting cards in fact does not matter even your computer is on a board and whatever software writing is you still have some firmware to boot load from some disk. Usually you would not be actually having a hard disk you would not be actually managing a hard disk here. Here you will be managing in fact the initial image which will be loaded will be decided by the boot loader. I think all of you do understand how a computer starts. So, there is always a bios and usually the initial whenever you power on a CPU there is a location to which it will be from which it will be loading the first instruction is always well specified it depends on the design of the processor. So, it will always go that. So, in that place usually that those address memory locations are being covered up by bios and now of course there is another variant which has in place of bios and this almost very genetic usually it is a non-resable kind of memory which is there. So, even if power fails anything happens you do not read or write into that. So, usually nowadays we use flash and remember when you are doing a flash upgrade because it is possible to do the upgrade of this ROMs and your computer itself can do it when it is running by switching but that is the risky thing because if you are doing at that time if the power goes something happens your machine is gone then because there is no way you can reboot it back actually unless you take that chip out and flash it again from somewhere else. So, laptops typically this problem is there the power should not fail you should fully charge before you do flash upgrade or ROM upgrade same thing is true even with your mobile phones. So, whenever you do android upgrade typically says that power should be full at that point of time power failure happens you will not be able to recoup. But most of the android phones have done a very smart thing we actually have a small bios which is never written what you actually flash is something which is after that. So, original boot loader is never rewritten. So, you can actually connect it to your PC and then again reflash it because for such a large number of people the number of phones will usually be larger than PCs and this kind of thing happens it will be a problem for company to manage and when you give a warranty of one year is going to be non viable thing. So, commercial things always has to be looked into. So, usually again this you can reboot do anything does not matter nothing will be lost is actually from where is flashed here it is always when you start it will take for some time to boot load actually the initial ROM will start boot loader code it will then say ok I have to it will mounted disk the disk driver will be there as part of the ROM the boot loader and then from there from certain location usually is a disk sector one cylinder one platter one on that it will start loading from there some particular code size based on that then you can read the file system and from that you can load the whole image and that image will be then control will be transferred you have the complete thing running ok after that here it won't be it's directly flashed on to this that is why we call it firmware firmware is never loaded again dynamically while your PC your kernel image is there in your slash boot if you see if you go into Linux file system slash boot you will have that kernel image you can actually maintain multiple kernel image when boot loader starts you can select the kernel image that's why you can have dual boot systems but there is also possible with mobile phones because they are also now being actually built in the same fashion you can actually have multiple OS running on the same mobile phone people have done that so that's a difference between firmware and software firmware is usually used inside so this is the opcode and then of course what else is required operation code has to operate on something some parameters so in this case usually it will be input port specification and output port specification and in the last because this can be corrupted by noise by error even if you are in a very small environment lens are not very large I always going to have a CRC check some systematic code will be there which can always be verified equivalent to parity bits actually and then of course in our case I have this kind of system I have multiple time switches so what will my input specification input port physical port which particular input thing and then time slot number there are two things which will be important so it usually it will be time switch number and when I am taking the input it can be even very well super multiplex time switch and I may have multiple PCM pulse code PCM streams multiple events so I have to specify even first of all the time switch because there is only one physical port within that what is the PCM stream code I am just putting additionally otherwise it is only one single even you simply can say input port and slot number so this is variable this has to be again depend on the switch there is no standard on this okay so time switch PCM stream number and then of course time slot physical port is time switch number remember time switch diagram so when I say time switch number I am giving port number but this might have multiple PCM streams coming that is why this additional field is being provided the same thing will repeat for the output if I have a single even suppose then you do not require a PCM stream number but the format will be general format need not be it is a vendor specific because this equipment this part of the exchange this is actually one of the board which will be sitting inside the exchange there are multiple boards connected through buses or the back plane so this is one of those boards actually which is there what is the PCL though pulse code modulation PCM stream even carrier is a PCM stream so similarly it will be done for this particular part for this output so that is the message which will be sent and this is which is fine and in response to that type 2 message will come do you have a class after this otherwise I will take five more minutes and it will be complete with the message type 2 message also so type 2 message is pretty simple and of course opcode is a 12 option so you require 4 bits here in this case type 2 message typically will again in the backward direction there is a separate message number this message number is not this in the forward direction the message will be numbered 1 2 3 4 5 in the reverse also similarly message numbers will be there if I receive 1 2 3 4 but 5 is not there 6 comes to 5 was lost so I can always now refer back and then send back the information for each message number so I can say this is the information which is depending on what was the opcode another thing and what I want to send back this structure will be decided by that and usually this is the reference message number so this information is about this message number which was received so when the reverse direction when this will be going this when this way this reference message number will be just simply copied from here if I am talking about this for example I say reserve the path from this port to this port with this thing this was received correctly reverse direction then I will copy this message number here put my own message number depending on what is whatever is my sequence I might put an information it was successful path has been reserved from this port to this port kind of thing some encoding will be there for the messages it is not a text message is encoded compressed message and then ultimately you will have a CRC some editing code has to be there error correction usually error correction usually is not used because you can always ask for a retransmission and this is always a response this is always a response the initiator is always this when it says there is a response this can keep on responding this can send multiple responses for every request reference number is there I can put same reference numbers and multiple responses see that is why this does not have a reference number to this message you have to understand that is why when it sends this message in response to this this will never do this is always the initiator other person is always the responder so the slave is this and master is this this is the reference message number is because of this that is a typical what we call pattern which will be there in master slave ok so I think next is the hardware part type 3 which is managed from here to switch so I think that is a very small thing so what we will do is we will discuss this thing in the next lecture and on Monday ok and if I can finish it before within that lecture then we will start with what we call input and output which is the performance of them we will now start basically with packet switching I want to create packet switch I will take simply a cross bar apply the packets at the input synchronize once and see how the input queuing or output queuing which one of them is better how to make an estimate some mathematics actually will be required I hope each one of you have already done especially movement generating functions and all that in some place in some course if it is not there I will give the reference and you can study sir here you assume that this is right cycle will there be any difference in configuration see there is a advantage because in type 3 thing I require less amount of lines if I use this you can do it other way around also but you will be investing you will be actually adding more copper on to your board your board size so your computer if you are going to make a cheaper one you will lose the market you try to reduce every wire every line which is there on the board as far as possible in your design you try to reduce path you try to reduce computation every bit counts because this is going to be replicated large number of times so you have to be cost conscious so cost conscious this always has to come from the day 1 when you start on your drawing board it cannot that you build up the design and later on you will start doing the cost cutting do not do cost cutting I say still cost optimization so engineering and science there is one major difference is science in science you never to look for optimization engineering means optimization reducing the cost so that it becomes viable commercially viable for more number of people otherwise it will only be can be used by the kings common people would not get the phone so we have been trying to reduce cost the whole telecom in fact whole networking which I have been telling is nothing but reducing cost by sharing sharing the links so anything which is dedicated to you I am trying to minimize on that that cost can go down and that is true I think in every engineering sector type 3 message type 3 is a hardware message being sent from here to the switch switching elements it cannot be structured because this is serial communication from here to here usually there will be serial line maybe parallel I do not it does not matter to me it can be USB so you will start from this bit and you will send it is a serial transmission of bit you have to do framing, deep framing everything will be involved in this case and then you will interpret that but from here it will be actually sent from the buses like there is a bus on which you will send in the back line you will be putting up the cards of this you will transmit the type 2 message for example so the message number will be the same in 3 cases only my reference number will be no reference number for example for this message number 1 was sent it resulted in generation of 3 type 2 messages all 3 type 2 messages will now be numbered 1, 2, 3 in the reverse direction this message number but they all should be referring to message number 1 so reference number should be common you are responding to which particular type 1 message you are telling that specifically because there is a slave it does not do anything on its own you send the command it responds you do not give the ask for anything so there is no pull it is everything is pushed so this guy other guy has to push the all commands all control actions