 So, basically why are we talking about wireless networks? This is something which is fairly obvious to most people, we want to access computing and communication services while on the move right. So, there are primarily three different types of such networks you know cellular networks which we are all used to the cell phones and so on. Then there is the wireless LANs which also operate in two modes, one is an infrastructure mode and one is an ad hoc mode. And off late there is a third type of such networks which are becoming popular which are called wireless MANs, which are again based somewhat on the wireless LAN technology, but they are expanded to longer ranges. So, given that let us do the first activity break here. So, let us say we are designing a wireless network. What are the factors that we need to consider in order to design a wireless network? So, this is the question. So, you can talk to your neighbors you can come up with some you know some idea of what do you think are the key factors that one needs to keep in mind while designing a wireless network and you know what we will try to see is how these factors affect the final engineering of the various systems. See theoretically it is very straight forward right. So, theoretically so let me try to give you an example here. So, theoretically we know that there is FDMA right, there is TDMA you may have heard these terms you may not have encountered the details it does not matter there is CSMA right all these things exist we know about all these various medium access control mechanisms right. So, the question we are asking ourselves now is why are there so many different types of mechanisms and when to use which one. So, what are the key factors for that one needs to keep in mind for designing wireless networks. So, in the case of wired networks it was pretty much straight forward right, we had by and large two options Ethernet and ATM and then most people today use Ethernet, so it is fairly straight forward. So, what is it that we need to keep in mind when we are designing wireless networks. So, what you can do is you can say the answer I will write it down then we will try to figure out which are the important ones. Bandwidth, number of users, somebody said location, what does that mean location, area, terrain, protocols, quality of service, bandwidth, security, power, mobility, cost somebody said cost. So, these are all mostly you know academic answers and what we will try to do is we will pick the one single one that you have to pick, suppose you have to pick one out of all this which one, you can pick one cost right, cost rules right, it is always cost finally, you see this is not science we are not doing it just for you know the beauty of the subject this technology and the whole idea is in giving users a service making profit. So, the cost is always the one which drives these technologies. So, if I had to pick one I would pick cost saying what is it that is determining the cost of this system, how do I keep my deployment cost the lowest, how do I increase my revenue and that is basically the whole game right. So, that is what we will see defines most of this work. So, the second question is so, this question second question that we are now answering is how does this cost impact the design of your wireless system, what is the impact of cost where is this cost come from. So, costs are hardware utilization of bandwidth, installation and maintenance, so the key cost here is licensing. So, if you have to pick one again it comes down to saying how much money am I paying before I can start operating right. So, it turns out that in wireless networks you have licensed bands and unlicensed bands right. So, for example, GSM is in the licensed band that is your cell phones 900 megahertz band right and you have Wi-Fi which is in the unlicensed band. It operates in the unlicensed spectrum 2.4 gigahertz is called the ISM band. So, Wi-Fi is designed to operate in the unlicensed band ok. So, just to give a better handle on these numbers can anybody tell me what is the licensing cost of for a GSM operator ok. Let us let us try to come up with some you know order of magnitude is it in 10000s lakhs crores crores right. So, that was everybody can kind of guess that it is likely to be in crores whatever is the number it is likely to be in crores. Now, what you are doing is basically it is about I think all the operators together in India pay about 1000 crores per year as licensing fees that is the numbers ok. So, all the operators put together. So, if you go to the telecom regulatory authorities of India website you will find these numbers there. So, now the question is how about Wi-Fi how much are we paying nothing right we are not paying anything for Wi-Fi whereas, in the case of GSM we are going to pay a lot of money ok. So, now let us go back and see a few more of these points before we come back to this question ok. So, now some generic things which are true. So, keep this in mind that there are two ways we need to go based on cost we have to think about ok I am paying so much money. So, I need to get the maximum out of the system so that I can recover my cost. In the other case I am not paying anything at all, but then I have some other goal in mind ok. So, keep these two things in mind and then let us go on to see what are the generic things which are common between all types of wireless networks be it GSM or be it Wi-Fi what are what is it that are that is common ok. What is common what is the key difference between wireless and wired again this some of these points you have already made no regulation of frequencies if I am in the license spectrum then I have to coordinate the usage of that spectrum right. In the wired case I do not have that problem at all right I can just run a wire from any machine to any machine as long as I have control over that physical space ok. So, another thing that we need to keep in mind is that the wireless medium typically has lower transmission rates right it generally has an order of magnitude or several orders of magnitude lower transmission rates the bandwidth is much less and the delays are much higher right. What are the delays several hundreds of milliseconds whereas, what would be the delay on an ethernet microseconds yeah I mean just about hundreds of microseconds ok. And also what happens is that we have higher loss rates in the wireless medium right it is susceptible to various types of interference. So, typically in a wired medium you have 1 bit in error in 10 power 6 bits right in the wireless medium what we have is about 1 bit in error in 10 power 3 bits ok. So, it is like almost a 1000 times more prone to errors as compared to a wired medium is that making sense. So, these are these things hold independent of whether I have any licensing cost to pay or whether I do not have anything to pay ok always shared medium many of you have mentioned that security is an important thing ok ok. So, then we also have these things about what are the other limitations that come into play you know there are some limitations because of the limited communication bandwidth we talked about you know limitations which are imposed because of mobility such as root breakages lack of mobility awareness. So, we will deal with all these things later on and finally, we have the limitations of the mobile device what are these limitations power and user interface form factor power and form factor are the two key limitations of the mobile device ok. So, now before we go on to this let us continue with our discussion ok. So, coming back here so, now we know that in wireless that the bandwidth is lower than the wired network the errors are more than the wired network and the delays are greater in the than the wired network ok. So, given that and given the fact that we are going to do all this you know we are going to take one of these two paths right either we have to go the licensing way or we have to go the unlicensed way let us try to see how do we design these networks ok. So, the question now is suppose we operate in the license spectrum suppose I am going to operate. So, suppose I have decided that I want to operate in the license spectrum what is my key consideration let us say I am given bandwidth 25 megahertz of bandwidth ok I am given 25 megahertz of bandwidth now what is my key consideration somebody said utilization right. So, the key thing here is I want to maximize utilization right what does that mean what does why do I want to maximize utilization is that basically what this means is I want to maximize number of simultaneous users right. So, number of simultaneous users is my call handling capacity right I want to maximize number of simultaneous users ok. I want to maximize the number of simultaneous users in the system right what other thing that I need to know access technology is what we are trying to build ok. So, assume that GSM does not exist this is 1990 target customer preferences right. So, the other thing we need to keep in mind is what is the target application ok what is the target application in this case it is voice see it is very important you can always design technology without keeping in mind what the application is for and then have the technology fail ok that is the key recipe for failure if you do not keep in mind who are the people who are going to use this and what are they going to use it for then that is a recipe for disaster right. So, GSM they did not do that. So, they were very clear in the beginning that their target application is voice and they want to maximize the number of simultaneous users in the system ok. So, you have 25 megahertz of bandwidth right. So, what do they do after that? So, can we take some decisions on the basis of this? So, we know that FDMA exists, TDMA exists ok all those things exist can we take some decisions now are we ready to say which way do we go? Given that my target application is voice how should I allocate the channel? What I am trying to get at is voice is basically periodic or what is called CBR traffic right. It is not exactly CBR, but it is ok CBR is a decent approximation ok. So, what this means is that I know that every 10 milliseconds I am going to get a voice packet from my voice code ok. So, that is a very important knowledge as far as the system designer is concerned I know that every 10 milliseconds moment of call is established every 10 milliseconds one packet is going to come from that flow that is a known thing right. The second question is of how do I maximize the utilization? How do I maximize the utilization? I have 25 megahertz of bandwidth multiplex right. So, I need to do some kind of a multiplexing mechanism. So, 25 megahertz of bandwidth I have to multiplex. So, what do I do? I have to do one frequency division multiplex access first ok. So, the first thing that you do is a frequency division multiple access which basically says that I will do 125 channels of 200 kilohertz each. See look at it this way suppose the entire 25 megahertz I were to give to one user right. What would happen? What is my capacity? Simultaneous. Simultaneous number of simultaneous calls capacity is? 1. So, correct if the whole 25 megahertz I give to one user my capacity is 1. So, the first thing that we do is we split it on the basis of frequency right. So, we now have 125 channels of 200 kilohertz each what is my capacity? 125 right, 125 is my capacity I want to further increase this right. So, even in the same frequency what we have to do now is we have to remember that even if I give this entire 200 kilohertz of frequency to a single user that user is going to have a packet coming in only every 10 milliseconds or so right. So, after that packet is transmitted the medium is going to be free is that making sense ok. So, the next thing that we do is after that we say the second round is I do a time division multiple axis I say 8 slots frame it is called a frame we can write channel 2 what that means is that I am going to put 8 users simultaneously in a frequency band. So, in that case it is the first user's turn, second user's turn, third user's turn so on 8th user's turn and then I come back to the first user's turn what is the capacity now ok. So, the capacity here is about 1000 users less than 1000 users actually approximate is that good enough how do we increase this capacity. So, we have just finished designing one part of the system which is saying what is the kind of multiple axis mechanism that we choose given the fact that we are going to pay a lot of money for this system ok. We need to try and maximize the utilization so we say that a FDMA based system followed by a TDMA based system is the most appropriate at this point right are we done can we increase the numbers can we why is it 200 kilo Hertz suppose I make it 100 kilo Hertz what happens to my capacity, my capacity will double right this is so obvious 200 kilo Hertz 125 channels if I make it 100 kilo Hertz I have 250 channels why do not I do 100 kilo Hertz quality of service what does that mean quality of service adjacent channel interference ok. So, the next thing here is when you look at why 200 kilo Hertz so what you see is you have the notion of your wave form right being like this let us say this is what you are saying as 200 kilo Hertz. So, the key is if you have so much of the signal available at the receivers end then the receiver is able to easily capture the signal right if we were to make it 100 kilo Hertz what would happen is there would be some overlap between one channel and the next ok. So, if we were to break it up so if we were to say that ok this is a 100 kilo Hertz signal so the next channel 100 kilo Hertz signal would also kind of begin here and go there right is that being sense this is how the various different channels are going to be spread out in time you are going to have channels one after the other each of them is going to have a peak at some point right. So, now around the peak you need a fair amount of the say signal in order to be able to capture it this is in you know kind of lame and terms that I am using here ok. So, if that is too small then if there is another interfering signal here for example then you are not going to be able to capture the signal effectively enough ok. So, what does this depend upon? So, this basically depends upon the modulation techniques signal processing techniques ok. So, depending upon the modulation techniques that you are using what is the signal processing techniques that you are using you can have different numbers for the 200 kilo Hertz ok. It has been found that 200 kilo Hertz is a reasonable reasonably good number see in technology there is always a trade off I can always do 100 kilo Hertz provided complexity increases complexity implies cost right always it comes back to the same thing if I increase complexity I am going to increase my cost because I then have to worry about maintenance I have to worry about a whole lot of other things right. So, 200 kilo Hertz is what the GSM standard decided at one point saying that this is what we will do ok. What else? How do we increase further? So, we know why the 200 kilo Hertz now. What else can I do? Choosing another access technique. So, remember that we are now in 1990 not 2006 in 1990 just before that we are trying to invent this technique right there is really not too many options available ok CDMA was not invented at that time this is history you know this is very interesting in technology history begins 10 years ago right yeah we can increase the slots per frame. So, the second number that we can try to play with in order to increase our number of users is to increase the slots per frame. So, suppose I were to make this 8 into 16 obviously again I can double the number of users that are there in my system correct why do not I do that correct. So, again I do not increase the number of slots per frame because if I were to go to 16 that means each user is going to get his turn once in 16 slots. So, this basically depends on what is the slot duration right ok. So, the slot duration is 577 microseconds alright each slot is 577 microseconds is it making sense what I am saying ok let us just try to draw one frame here. So, what we have is we have ok that is the way it goes alright. So, each slot is 577 microseconds that is the width of every slot this is in each frequency alright. So, for each channel you have these 8 slots each of 577 microseconds ok now the question of course, is there why 577 microseconds right. Obviously, if we make it smaller we can fit more people see that is the way technology works. So, we will deal with why 577 microseconds in some time first let us see given that it is 577 microseconds why is it that we have 8 slots per frame can we see. So, how often do I get a turn when it is 577 microseconds 8 into 577 which is right. So, which is about the time that my see remember that I have a voice packet right. So, my voice arrival is 260 and 10 milliseconds ok the we will do the bits calculation later basically what we know is one voice packet is coming every 10 milliseconds right. So, we need a slot for that voice packet to go out every 10 milliseconds or so ok. So, that is why it has been found that if you go to 16 that becomes too far apart that means your voice packets start getting queued up ok. So, which is why we stick with 200 kilohertz and 8 slots alright. What else do we need to think of reuse correct. So, another aspect of the system is how often can we reuse. So, now what we have is we have what is called a cell which we imagine as a circle, but in reality for modeling purposes this is modeled as a hexagon. So, this we are saying is a cell right. So, within a cell we have about so we are saying approximately 992 users ok. So, 1000 is generally some channels are used for control and so on. So, we have about 992 users. So, now what we want to do is we want to increase the number of these users correct. What do we do? We have to we have to replicate the cell see right now we have not got into real location of frequency yet. So, what we are saying is this is a cell and if I want let us say I am going to operate in an area like this then I will put one cell there I will put one cell here and I am done right. So, what is the problem if I do that let us say I put a cell here I put another cell here this can support another 992 users. So, in total this area this is my area of operation I can support twice the number of users ok. What is the problem if I do this interference. So, the problem is if you look at users who are in this area if they are using the same frequency channel talking to one base station here and another user is using the same frequency channel to talk to another base station here what happens is they are going to interfere right. So, we have to come up with a mechanism by which we can do this in an interference free manner right. So, how do we do that? We say that let us draw these hexagons ok. So, whole bunch of hexagons are drawn and it is not each hexagon that uses all the cells, but a group of hexagons together use all the frequencies ok. So, what happens here is group of hexagons that is basically what happens. So, suppose we take a 4 cell reuse pattern then you know that if I am going to use a certain set of frequencies in this cell then you know that the same set of frequencies can be reused in this cell ok. If you take a 7 cell reuse pattern, so these are called reuse patterns. If you take a 7 cell reuse patterns again there is a way of calculating which is the other hexagon in which the same set of frequencies can be reused ok. So, we do not need to go into the details of that calculation because it basically is something like a night move on a chess board ok. So, you go 2 steps in one direction you go one step in another direction then you can find the other cell in which the same set of frequencies that can be used ok. So, if this is a 4 cell reuse pattern this is how you get to the other one. If it is a 7 cell reuse pattern you would go 2 cells in one direction and then one cell in the other direction ok. So, how that comes about and all we have to get into hexagonal geometry to arrive at that. So, right now let us not worry about that what basically it means is that a group of cells once you take a group of cells which are together going to utilize the same all the frequencies this pattern now can be repeated indefinitely right. So, this is what happens inside a cell. So, let us do one small exercise ok. So, let us say we have total bandwidth just to make sure that we understand what we are talking about. Let us say the total bandwidth is 33 megahertz ok and we have 33 megahertz of bandwidth available totally and we are using 25 kilohertz per connection. So, what is this into 2 that is shown here? Duplex. Right. So, it is a voice connection. So, we are going to use 25 kilohertz in one direction and 25 kilohertz in the other direction because it is voice it is a full duplex connection right. You know that one packet is going to go this way and you have to make the same allocation for a packet going that way correct. So, how many channels are available for a 4 cell reuse pattern ok how do you compute this 165 so, we know the magic number how do we get to the magic number So, what we compute is the total number of channels per cluster how do you get to that? 33 into 10 power 6 divided by 25 into 10 power 3 into 2 right is that correct that comes to 660. So, I know that I have total number of channels per cluster that is available to me is 660 correct. So, if I want to do 4 cell reuse 4 cell reuse is nothing, but 660 divided by 4 that is the number of channels that are available in each cell of a 4 cell reuse pattern ok. The 7 cell is a little bit tricky it is not very straight forward 660 divided by 7 ok 660 divided by 4 turned out to be a nice number like 165 660 divided by 7 turns out to be 94 point something right what does it come out to 94.4 correct. So, what do you do with the 0.4 you do not have a 0.4 channel correct. Round off to the lower number. So, most of the cells will round off to the lower number, but 1 cell will round off to the higher number ok. So, that is what happens when you do not because the total number is important 660 is the important number in the cluster I have 660 channels right. So, if it comes out to be 94.4 then I say most of the most of my cells in the cluster are going to be doing 94, but whatever is left over the last cell is going to do 94 plus whatever is the additional that you get by adding up all the 0.4 is that mean sense otherwise if you just did 94 what would happen you would be left with 94 into 7 channels you would lose a certain number of channels ok that mean sense ok. So, so far what we have been able to do is we have been able to figure out what kind of allocation mechanism that we do how do we go about deciding how many cells that we need ok. So, most of left which is about deciding you know what is the kind of transmission mechanism and all that. So, we will come to that later on. So, right now let me continue with just going back to the overview. So, this is where we had stopped right. So, this is again something which we need to remember that in the physical layer we have seen that transmission errors and interference are the two main things ok. So, the why have I put this slide the reason for that is that periodically we will be revisiting this to see how things change from the wired network to the wireless network ok. So, in the wired network we know that the transmission errors are low the interference is less. So, we have gone for a different physical layer in the wired versus the wireless ok. So, again in the link layer there are issues with media access right. So, in the wired case is there an issue with media access? No, is there any issue with media access? Suppose I am on a LAN yes. Suppose it is a switch network ok. Suppose I am on a switch LAN which is what most of us have today then it is ok there is it is almost like a one to one communication whereas, in the wireless case we see that accessing the medium how to determine that somebody else is also accessing the medium is a fairly important thing correct. What is this handoff? When I go from one area to another right because it is a wireless implies mobility often wireless implies that you could also be mobile you know. If it is fixed wireless then typically it is made explicit that this is a fixed wireless system otherwise wireless typically implies mobility ok. Similarly when we go up to the network layer so, all these things we are going to see before the end of this course. When we go up to the network layer again we have issues like addressing suppose one guy has moved from this subnet into that subnet how do we change the addressing mechanism how do we ensure that the packets get routed accordingly right. At the transport layer again we have issues of figuring out what is the congestion control mechanism can we use the same mechanism which holds in the wired network and just applied directly into the wireless network. So, the point of making this is applications of course, we know so, the point of saying this is that the emphasis is always going to be that is the way technology evolves the emphasis is always on trying to understand how do I take what is existing and push it to the new system that I want to build right. So, ethernet exist I want to see can I use ethernet in order to bring it to the wireless medium right TCP exists for wired networks I want to see can I use TCP again in wireless. So, it is always the attempt is always to say that I already know how this works how do I take it into the wireless domain is it usable as it is if so, my job is done if not that is when I need to invent something else ok. So, all this stuff much of this we have seen this is kind of a recap cellular systems the basic idea is that you have a single hop wireless connectivity and then mobile hosts can change cells while they are communicating handoffs occur when the mobile hosts starts communicating via a new base station correct ok. What are the factors that determine the cell size again that we have already gone through this is just the recap number of users is the key factor and the multiplexing and the transmission technologies ok. For example, if we are using CDMA we have a notion of what is called a soft handoff soft side then what else we have a limited number of frequencies which implies that we have a limited number of channels if we use a very high power antenna then we are going to have a larger cell size which will limit the total number of users that I have in my system. So, smaller cell size will give us more frequency reuse ok. So, the base stations implement space division multiplex cluster is a group of nearby base stations that together use all the available channels ok. Mobile stations communicate only via the base station and as demand increases the number of base stations can be increased the transmitter power is decreased accordingly ok. Any doubts on this ok. So, now what we have kind of understood is what goes on inside a cell I mean we cannot claim to have fully understood it, but we have an idea of what goes on inside a cell ok. So, now what we want to see is how does the rest of the system operate yeah ok. So, let us come to the other next aspect of it ok. So, take a look at this slide and we have these boxes called HLR, VLR, MSC and so on ok. So, what we are trying to understand now is what is the purpose of all these other boxes ok. So, we have understood how a base station communicates with a mobile in itself right. What else is required to make the system go? What else is required? Suppose again we are back in 1990 we have to start inventing this right. What else is required? Connectivity to the base station is there. So, the mobile is able to connect to the base station. So, now think of it like this. I want to make a call from my mobile to let us say another mobile that is one thing. Another thing that I want to do is I want to make a call from my mobile to a PSTN number ok. What are the other components that I would need in my system to make this happen? Switching between wireless and PSTN, track the location of the user or the called party, caller and callee. Authentication, billing, database for traffic analysis, switching system ok. Think of it this way. One mobile wants to call another mobile, one mobile wants to call PSTN ok. Given that do we have all the components? So, that is signaling between wireless and PSTN signaling and encoding ok. Connectivity with other cellular systems PSTN and other cellular systems ok. So, let us look at it this way. Suppose we are now designing this system right. Let us say we have. So, I am going to just draw some vague shapes. So, let us say you have a mobile in this cell ok. So, you have a mobile 1 which is talking through BS 1 ok. Think of it like this and I have a mobile 2 which is talking through BS 2 ok. So, now, I want this mobile wants to make a call to the other mobile right. How do I connect both of these guys that is the question correct. Can I connect it like this? Why? So, the answer is that this is wrong. It should route through some switching center this is correct ok. Why is this correct and this wrong? Because both are moving. Can't I locate? If I have if all the BS's are connected if I have 10 BS's all the BS's are connected to each other. Can I not locate where the mobile is? I can. Database can be maintained at the switch. So, what we are trying to do again is to build the system bottom up see the less number of components that we have to deal with the easier our life. So, the first question that we are asking ourselves is can we just do it using only the base stations. Suppose, I take a wire from each base station and connected to every other base station in my cluster or my cell will it do. So, suppose I am willing to live with the number of links suppose I have a very fast system for locating the mobile ok. I have a very quick way of determining I just query all the base stations ask them you know is this mobile in your network in your cell or not the base station can reply everything is very fast. PSTN is missing suppose I do not want to talk to PSTN networks. Now tell me billing somebody said billing that is the key right. If I have all the base stations talking to each other where does operator make money right. How does he know who is calling how long he is calling that is going to be the key problem right. Billing is the key thing that they want to make it easy. So, right now so even if you are in the adjacent base station. So, typically what will happen is even if mobile 1 wants to call mobile 3 in the same base station actually I should have asked this question that would have made it more easy right. Even if mobile 1 wants to call mobile 3 the call is not rooted just through this base station it has to go to the switch and come back primarily because this switch is handling billing right and location both these things are handled by the switch ok. How many calls can such a switch handle right. Can I connect up all the all the base stations in mobile to a single switch I will kill it right if I do that. So, what is done is we say that let us separate out some of these things or let me ask the question differently this switch is handling billing and location right. Where is the frequency allocation being done base station is allocating frequencies to the mobile correct, who is allocating the frequency to the base station saying that you use these 5 channels you use the 4 channels switch right. So, this switch actually if you see has you know two distinct set of functions one is to interface with the external world one is to do all this billing location tracking mechanism which is actually independent of managing the radio resources is that making sense ok. So, I need a switch which is also managing my radio resources ok and because these numbers are going to explode what you have is something like this you have your cell here you have the base stations which connect to what is called the BSc this is a smaller guy right smaller degree. So, remember it in terms of the same degrees right which connects to a BSc and you have a bunch of BSc's which connects to your MSC ok. So, this guy does the billing and location management while the BSc's do the radio resource management again technically there is no harm in putting all of them in one huge box is just going to increase the complexity is just going to increase your propensity to failure right. So, now with BSc's each BSc basically controls a cluster. So, you have kind of done a divide and conquer on the whole problem instead of having the MSC control all the base stations in the system what you are saying is I will have BSc's each BSc which will control each cluster and then the MSC controls all the BSc's together and that is how the routing and the call handling mechanism takes place ok. So, that is why you see all these things. So, in this figure I have kind of simplified it to show that this just MSC they are connected directly to the MSC there are these two small boxes shown as HLR and VLR right what do they stand for? So, the MSC has talks to somebody called the home location register and the visitor location register what is the purpose of these guys? They help us to do the location management of the system ok. What may be going into each of these we will come to that in a moment ok. So, the MSC's are connected to other MSC's and it is also connected an MSC also could be connected to PSTN. So, typically you have what is called a gateway MSC GMSC per operator which connects to other networks ok. So, now the question is when we are doing handoff. So, we will just do this one part and then we will do one small activity ok. So, how do you do a handoff? What is a handoff? When the mobile node is moving from one cell area of one cell to the area of another cell ok I may be doing handoff ok. On what basis do I do this handoff power right it depends upon how far away I am from that base station if I am near to that base station I do not need to do a handoff if I am far away and if I am in the boundary then I have to do the handoff right. How do I do that? By restraining so, basically it is like what I am saying here see there is one way of doing it is that the base station keeps track of all the mobiles right. Base station knows that I am receiving this mobile at this received power level you know this mobile. So, the power level has a correspondence with the distance right. So, the base station can do all the calculation and then it can say that you switch from this base station to another base station ok that is one way of doing it ok. Another way of doing it is the mobile takes charge of it mobile takes the responsibility of doing the handoff right which one is easier? The mobile deciding mobile does not really decide, but the mobile does all the work see the decision who takes the decision? No. BSE. BSE. See always we are talking about within a cluster right. So, the BSE is the one which takes the decision right the MSE is not does not care about radio resource management at all the MSE is simply going to say do I have enough information to establish the call do I have enough information to build the call that is all the MSE cares about it is the BSE which says do I have enough resources to make sure that this call continues ok. So, the BSE is the one which is doing that. So, the decision is taken at the BSE, but what typically happens is that the mobile at any given point of time is measuring about you know 6 base stations the signal strengths from 6 different base stations in that area ok and it periodically keeps transmitting the received signal strength information to the base station ok which it is communicating that in turn goes to the BSE the BSE looks at all the numbers and at some point it figures out that I need to shift this mobile from this base station to that base station ok. So, that is called a handoff right and the third type of handoff is when the mobile moves across areas controlled by different MSE. So, we will just come back to this in a moment ok. So, let us do the handoff business ok. So, what we are saying is let us look at our system again we have our cell we have mobile 1 here which is connected through BSE 1 here which is connected to a BSE 1 which is connected to an MSE right and MSE is kind of talking to 2 databases which is the HLR and the VLR ok. Let us forget about the VLR for now ok. So, this mobile is let us say moving from this cell to another cell ok. So, when this mobile is moving from one cell to the other cell what are the activities that take place? So, yeah it has to get a frequency channel in this additional adjacent cell right it has to get a frequency channel here and it has to notify this base station that I am switching from here to here right. Does any change need to be done in the HLR ok? What is the HLR? What information is being stored in the HLR? So, home location register means what HLR basically stores information like you know authentication ok. Is this mobile roaming enabled you know should I services that are subscribed for ok authentication information services information and the location. Location in some sense ok what sense is what we are trying to now find out ok. So, the mobile is moving from here to here and should I update the HLR that is the question yes no ok. Let me ask the question differently what are the different strategies by which we can do the location update see in order to understand why we need HLR and VLR as two different entities let us try to ask this question you know what are the different ways how often can I what are the ways in which I can do location management. Suppose there is an incoming call suppose there is a call that is coming into this MSC. This MSC needs to locate where the mobile is ok suppose I have only two cells I have one MSC one BSE and two cells ok. I need to locate where the mobile is what are the ways in which I can do that. I can just ask all the cells right I can just go and ask all the cells is this mobile there in your cell get an answer and then I can do the allocation I can do the call establishment. Do I need HLR for that I do not need a HLR right I just do a direct query on the network find out where is the mobile and then I can I look at I can establish the call in that manner right what is the drawback of that time delay no bandwidth this is all wired network see after the base station they are all wired networks ok. So, bandwidth is there only at the wireless interface that is not a constraint after that ok. So, the key thing is as the number of cells in my system grows I am not going to scale no if I have to page all the cell that is called paging. If I have to page in all the cells to find out is this mobile there in your cell or in your cell or in your cell then I am going to take too long before I can locate the mobile one and secondly I may consume too much of resources lot of messages will go back and forth to figure out where is this mobile. So, that is why the HLR is used right. So, let us do one small exercise. So, consider a network where we have the geographical area divided into let us say roughly square cells ok. So, let us say you have the geographical area divided into 2 dimensional array of square cells ok. So, now let us say you have this is my area of operation and I have an MSC which is controlling this area ok. Now, when there is an incoming call ok I have to find the location of the mobile. So, the mobile could be in any of these cells correct. So, the mobile could be in any of these cells I need to find the location of these mobiles ok. What are the various ways in which I can do this ok. What we are trying to understand here is what is the trade-off involved in this location update. So, one strategy of location update is mobile never updates mobile never informs updates whatever location register HLR let me write HLR along with the MSC I have an HLR. So, one strategy is the mobile just does not bother to do anything. So, what happens here corresponding to this strategy what is the action that has to be taken in order to do the location I have to page all the cells right. So, in this case I have to page all the cells the other strategy is mobile updates each time it changes a cell right that is the other extreme you have to first we have to figure out the two extremes right where we are operating. One extreme is the mobile does not say anything at all the other extreme is the mobile says every time it moves a cell it sends a message to the MSC saying that update the HLR this is the current location for me correct. So, here what is the cost involved I have to send a message, but I have only one look up. So, the MSC knows that if I do a look up I am going to be able to find this mobile correct. So, if I look up of the HLR I am going to be able to find this mobile ok. What is the cost involved here the number of messages that I send you know if I am just moving around imagine like this I am not making a call I am just moving around ok. Even if I am just moving around or even if I am just you know kind of going from here to here across these two boundaries right maybe this building is covered by one base station maybe the next building is covered by another base station and I keep going to the next building for chat right. So, I keep going back and forth back and forth and then these messages this network has to keep sending these messages to make sure that it can always find me right. So, that is another cost which is associated with this strategy. So, what is the golden mean right. So, the mobile updates periodically. So, either you can set up periodically as in time periodically or you can set up periodically as in movement periodically right. You can say that I will update my location every one hour that is time periodic or you can say that I will update my location every time I have moved three cells ok. After I move three cells I will update my location after every three cells that I move I will update my location ok. So, this is this could be over time or this could be over number of cells ok. So, now the question is number of cells to be paged how does the system find out when the mobile updates periodically. Suppose I am going to update every three cells every three moves let me say moves ok. So, this is the question from one of my exams is the question making sense what we are saying is now I have decided a strategy. So, let us say this three I have decided upon a strategy that I will update the HLR every three cells that I move ok. So, now how many cells does the system need to page in order to find me is it making sense. See I may have moved three cells in any direction ok I could have moved like that I could have moved like this I could have moved in any direction. So, you have to figure out how many cells does the system need to page the system does not know in which direction I have moved right because the till the update comes the system does not know where I am. Suppose at that point there is an incoming call how many cells does the system have to page to locate me one BS is controlling one cell right. So, all these dots that I have shown are BSs how many cells do I need to move do I need to find 20 how do you get that right you can count it. So, I see basically we are saying that the mobile updates every three moves right. So, when you say that the mobile updates every three moves suppose let us say it is moving in this direction. So, this is one move this is second move and when it makes the third move when it reaches here it is going to give an update correct. So, once the if so, I know when the mobile is here when the mobile reaches here also I will know I will get an update correct. So, what I have to search for is the intermediate nodes right how many of them will be there. So, if I am going in this direction I have 3 if I am going in that direction I will have 3 and I could be going in that direction right 3 square basically that means sense. So, this will basically work out to be no not 3 square I could I could go in this direction I could go in this direction correct. So, this is 36 right if I go 3 more cells here it is the same thing no one move second move third move at this move I am going to update. So, I have 1 2 3 4 5 5 square yeah. So, I have 5 square 25 cells to be paged is that being sense let me just draw this part also. So, what we are saying is that when the mobile reaches this cell this cell marked as x it starts from this cell marked as o when it reaches this cell marked as x is going to give an update to the location register under this strategy or it could reach this cell marked as x again it would give an update the same way when it reaches somewhere here and somewhere here it is going to give an update. So, that is the total number of cells that you need to page in order to be able to find out in order to locate the mobile when it is using a strategy of updating every 3 moves. Can we generalize this suppose it is k k moves what will be the expression 2 k minus 1 right. So, it will be 2 k minus 1 whole square that will be the number of cells that you need to page in order to locate the mobile correct. So, that was just an exercise to find out how intense this problem of location update is. So, when you talk about it in theory that it sounds that it is not a big deal 3 sounds like a good number you know updating every 3 every 3 moves right, but every 3 moves itself entails a paging of about 25 cells. So, let me just do this last one and then we can stop. So, this is basically a fairly dry slide, but it is still required because it is information. So, the first generation systems were analog then you had second generation systems which were GSM right. So, the important thing to note here is the data rates GSM had less than 9.6 kbps data rates primarily because it was a circuit switch system. So, we will be studying it in detail soon then you have the 2.5 g systems which is GPRS is an example you go from GSM to GPRS which gives you about 115 kbps in theory in practice it is about 50 kbps. So, we will see how to take the GSM system and what is the change that is required in the GSM system in order to go to GPRS and then you have 3G systems some of which are seeing deployment now which are supposed to be you know high speed data access systems. How many of you are hungry? Let us go. So, we will take a break now we will come back after lunch.