Lecture 46 : 5G SRS design – part II

Okay, welcome to the class and we have been discussing sounding reference signals. So, we discussed three use cases of sounding reference signal and started discussing its design as well. Okay, so we will continue discussing this, this will take us some time to finish and hopefully we will do it in coming few classes. So, what did we discuss about sounding reference signal? This is a quick recapture of that, okay and one thing we discussed was this SRS is limited to a maximum of four antenna ports. So, I will be referring to sounding reference signal as SRS, okay. So, sounding reference signal is very long, SRS is very small, okay, so it is limited to a maximum of four antenna ports and it is, it has basically a QAM structure, okay and the reason it is called QAM because if you look at it, these are the brushes of the QAM and then there is a gap in between, right, okay and so it is limited to four, maximum of four antenna port which can be sounded together and now this is, this time duration of SRS and in terms of symbols, okay and it can be 1, 2 or 4 consecutive OFDM symbols within a slot, okay and if you look at it, how many consecutive symbols are here? There is only one symbol and here it is 4 SRS symbols. We will see today why these four consecutive SRS symbols are there, okay. Why do you need to repeat the symbols like this, okay? So, the reason will become clear in today's class, okay and now the starting symbol duration is L0, okay, it starts from some symbol, okay and goes up to like, continues up to 1, 2 or 4 symbols after that consecutively, okay. Now the thing here is we discussed it in the last class that SRS comes only at the end of the slot, okay, the reason being that is like the first, the beginning of the slot is actually occupied by the downlink, okay. So, if it is a TDD system, it will switch from downlink to uplink, okay and you will see, I mean that is where typically it comes, okay, should keep that in mind, okay. Now before we get into different properties of this SRS, okay, so the standard defines three different kind of time domain structures, okay and that is what is we should understand, okay. One, so periodic, aperiodic and semi-persistent, so all this will become clear once I show you this picture, okay, so we will go through all of that in detail, okay, so the first one shows periodic SRS transmission, second one shows semi-persistent transmission and third one shows aperiodic transmission, okay. It is crucial to understand this to get a complete picture of this SRS transmission, so what is happening here is, if you look at it, so this, the first one is basically the periodic SRS transmission and if you look, so what is happening here, like what am I saying here, okay, so there is this message which comes from the higher layer, higher layer has to get into configuring the sounding reference signal which is basically the RFC, okay. So, now this message, okay, physical layer is not involved in configuring this periodic SRS transmission, okay, so what it says here is, I am talking about the configuration, of course the signals are transmitted by the physical layer, but who configures it, okay, right, so each layer has its own part, okay, so we are getting into link, we get into other layer very few times, okay, just one example we spoke about scheduler, okay, which does not sit in the physical layer, but it controls the physical layer. Similarly, here the RFC for periodic SRS transmission, it gets into picture and it sends a message to the user that the SRS is being configured with this much frequency, okay, and then it continues after that, okay, that UE will keep transmitting SRS continuously, okay, and unless and until basically the RFC tells that to stop transmitting, okay, so the message to stop for this transmitting this again has to come from RFC, so now message comes and UE starts transmitting the SRS, okay, so this is periodic and periodic of course like it has to be transmitted with a certain period, okay, but it continues, okay. Now, what happens is there and then this is the first one, okay, then this is called semi-persistent, okay, and semi-persistent basically is something similar to periodic, okay, but what happens is that first this RFC transmission, this RFC will get into picture saying that, okay, this is you have to transmit the SRS, okay, correct, so it will be configured by SRS is always configured by RFC, okay. Now, in the semi-persistent transmission, you see the difference, okay, MAC gets into picture, okay, so there is this MAC control element activation, so it will be configured by RFC, okay, you see then here but it is activated by the MAC, control information is sent by the MAC layer saying that please activate it for this much period, okay, or please activate it, okay. Now, then after that again the control element is sent by the MAC layer to deactivate it, okay, right, so here it MAC does not get into picture and it continues, so this RFC configures it and it continues, right, okay, here it will be explicitly, okay, the semi-persistent, okay, it says that MAC will send this control information, this SRS transmission will start, then after that MAC again says that please stop transmitting, okay, so then it is similar to a periodic, similar to periodic but MAC says that transmit and then MAC only says that stop, okay. So, this is basically see what is happening, there is a higher layer, there is this layer sits above MAC, okay, so lot more many more delays are there, okay, because physical layer then MAC layer, then RFC, okay, so if the signaling is coming from RFC layer, lot more delay are involved because UE has to go up to the RFC layer to read this message and configure it, okay, so this is basic and this is has the highest configuration delay, here it is basically is the lowest configuration delay, lower configuration delay than this, then you look at this a periodic transmission, okay, what is happening? So, this SRS has been configured, okay, then you look at where is this coming from? This coming from the DCI, the physical layer gets involved, okay, here it is, so SRS has been configured, whenever the base station wants the UE to transmit the SRS, it sends this, this is indicated in the DCI itself, okay, and then for that duration the SRS is transmitted, done, that is all, again when this SRS is required by the user to transmit, then again the DCI is transmitted and it is indicated using the DCI, okay. Now, what are the, so this is three different kind of configurations, now why this, why are three different kind of configurations or three different kind of transmission? Let us look at this, okay, it will become clear like why this a periodic is required if you understand this, okay, let us say user is in inactive mode, okay, basically it is not transmitting and receiving data, it is connected to the network, okay, so I am using these terms, okay, so what is like, what do we mean by like it is connecting to the network, basically it is getting, it is locked to the network, getting some control information, okay, but not actively receiving or transmitting data, okay, this is some and then base station has suddenly received some data for this user, okay, then what happens is and or user has to transmit some data in the uplink, then base station does not have the uplink channel information, so what will happen, base station will send a, it will configure this a periodic transmission for a very small duration to get the uplink channel information, okay, right. So, and it is, so the data is such that it is like limited amount of data, it has come suddenly and you need to transmit, okay, so but base station does not have the channel, uplink channel information, so it configures this SRS, so that it can get the channel information, okay, right, so this is one use case of this a periodic transmission, wherein base station needs the uplink channel information urgently, okay, right, let us say URLC transmission, it has come and then you need to transmit, then this kind of like sudden requirement for uplink channel information is there, then this a periodic SRS will be configured, okay. And then this is other extreme, wherein user is continuously transmitting data, it is a heavy file which needs to be uploaded or which needs to be downloaded, okay, so and continuous exchange of information is happening and then base station continuously need the uplink channel information. So, it configures it and then forgets it for quite some time, okay and because this has to happen for a very long time, okay, MAC does not get into picture, neither physical layer gets into picture, okay, so it is basically the urgency of the SRS or the urgency of the channel information based on which the three different kind of SRS transmissions are configured, this is very urgent, something in between, this is like it is going on continuously, so the message exchange is to configure is very few, okay, it is only RRC, here MAC gets involved, here it is happening at the physical layer level, okay, so whenever SRS needs to be transmitted, DCI needs to be transmitted before that which indicates where SRS has to be transmitted for how much long duration, okay, right. So, these are three different time domain structures, fine, okay, so that was the time domain structure, okay, now we are going to understand this frequency domain structure of SRS, okay, so what it, you look at this bullet, what is this bullet telling that which portion of a bandwidth part needs to be sounded by an SRS resource, okay, so we understand this notion of bandwidth part, that is a part of the bandwidth for which UE is configured to transmit its data, okay, and we have discussed it without like mathematically defining it, okay, but this is the notion of a bandwidth part, okay. Now, what this bullet says is that UE is supposed to be like configured for this bandwidth part and if it is like high end UE, it is configured for whole 100 megahertz bandwidth, okay, if it is like cheaper or like IoT kind of device, it is configured only for a very small number of small bandwidth part, okay, so what we are discussing here is, okay, so this SRS and frequency domain, okay, base station might be interested in estimating the small part of that bandwidth part or the complete bandwidth part, okay, right, and it could happen that when the UE is basically entering the network, base station wants to find out the channel of the complete bandwidth part because it does not have any information, right, okay, UE is already into the network and base station has already done that like already estimated the complete channel and it knows that the channel of this user is bad in this bandwidth part, so it will not touch this bandwidth part, it will only want to find out the channel, updated channel information of a part of the bandwidth part, okay, right, because it already knows that the part of the bandwidth part has a bad channel, correct, okay, so there are different configurations, there are different possibilities starting from the full bandwidth part to a part of the bandwidth part, right, the channel needs to be estimated by the base station in the uplink, okay, that is what we are saying here, okay, this bullet is look simple but it is heavily loaded, okay, now so what are this, this picture will make it clear, okay, so what else, this is the bandwidth part, so user only understand that this is where the 5G system is, okay, right, so now what are we saying here, so this is the bandwidth part, complete bandwidth part, what are we saying here, so this is the maximum sounding bandwidth, okay, so base station wants to figure out what is the channel here in this bandwidth part, it could be complete bandwidth part also, okay, and if I basically say this that if I stretch these two lines from here to here, it becomes the complete bandwidth part, okay, but for this example what we are saying here is that base station wants to find out the channel in this bandwidth part, okay, so and we call this the maximum sounding bandwidth and this is denoted as MSRS0, okay, this all these notations will keep coming repeatedly, okay, so as and when they come will again stress upon that, okay, so what is this MSRS0? This number tells you the maximum sounding bandwidth in a bandwidth part, okay, right and it is of course specified in terms of PRBs, okay, fine and now so this is basically this, the maximum sounding bandwidth, you look at this, okay, shift, what is this end shift if you will, the way it is indicated it becomes clear what it means, okay, shift is basically from the beginning of the bandwidth part from where you should start sounding, start sending the SRS, so you should, this is the start of the sounding bandwidth, so base station wants to find out the channel in this part of the bandwidth part, okay, right, so this is the start, end shift is the start, fine, okay, end shift again how is it specified will come back to it, okay, but this is the start and this is the maximum bandwidth which needs to be sounded, is this clear, okay, so what does that mean? That user needs to send SRS in this part and we know the structure, the QAM structure you remember, okay, so we will send the QAM structure in this part, okay, but in how do you configure where should that SRS be sent that is this is a start and then we will be told that okay, please send SRS for these many PRDs, but these are given a name, okay, this is a start, shift is the end shift is a start and this is the maximum sounding bandwidth, okay, this actual sounding bandwidth B hop and NRRC, I will come back to it, okay, once we start talking about this hopping all that, we will basically, we will then this B hop and NRRC will become clear at that stage, okay, I will not get into the details of it, they are there, so we will live with it, okay, but we will come back and explain that, at this stage what are we saying? This is a start, this is a complete bandwidth part and UE needs to send SRS only in this part of the bandwidth path, right, okay, because base station is interested only in estimating the channel here, so the picture should be in our mind is that UE will be sending that QAM structure in this path, correct, okay, but UE needs to be told where to start and how many PRDs to send, so that is how, but these are the variables which are used in the standard to define the start and then this maximum sounding bandwidth, okay, how is it sounded? All this will get involved, but this picture should be clear, this is a table which is defined in the standard, do not look at this, I will tell you what it means, okay, so this is a 64 row table, okay, and I have shown only a part of that, because it becomes difficult to like bring in the whole table, so 46 rows I have shown, okay, now what it shows here is, so first column is difficult to read here, but this is CSRS, okay, fine, and this parameter CSRS will run from 0 to 63, because it has 64 rows, okay, correct, this table has 64 rows and it selects SRS bandwidth from a table, so this parameter will select the SRS bandwidth over which the SRS needs to be transmitted, fine, okay, so smallest SRS transmission bandwidth is 4 RVs, you look at this, what did we learn about MSRS 0? This is the part of the bandwidth which needs to be sounded, okay, over which the SRS needs to be transmitted, so the maximum, so the minimum bandwidth over which the SRS can be transmitted is 4, okay, so this is the first row, okay, index by 0, and the maximum is 272 RVs, okay, so UE can send SRS over 272 RVs in one shot, so the con structure, correct, okay, and how will it be indicated to the UE that it has to be transmitted, it is indicated using CSRS, and the table is known, table is stored in both the link, base station and UE memory, correct, okay, now if it is DCI based, so this CSRS information has to come in DCI, correct, so this is otherwise MAC and RRC has to get into picture, okay, but this is a table, okay, and through this table only the sounding, maximum sounding bandwidth is indicated, okay, fine, so and this is a table in the standard, okay, so this is a reference, and will, so using this, okay, how is it configured, I will show you an example, okay, right, so this is a broad outline of link, how the CSRS, this parameter is used, let us take an example, okay, fine, so what happens is, we consider this example CSRS is 13, I am looking at this particular row, okay, so I am zooming in, I am writing it over there, that is all, okay, so this is that row, fine, so that is what I said, table becomes too small, okay, so what are we saying here, so this CSRS is 13, MSRS 0, what is this, this is a maximum sounding bandwidth in terms of PRBs, so base station wants to sound out 48 PRBs, which 48 PRBs that is indicated by n shift, correct, okay, right, now these different columns will understand what is happening here, okay, there is a SRS configuration exercise that is what we are going to do, okay, so what is the first parameter, what did we say, this is familiar to us, okay, what is this, this MSRS 0 determines the maximum bandwidth within the bandwidth path that can be sounded out by the configured SRS, right, okay, so base station is if DSRS, so MSRS 0 is basically the maximum bandwidth, okay, which needs to be sounded out, fine, so this look at this number, okay, right, do not look at any other number will slowly decipher all of them, okay, so this number alone, fine, okay, now what is this, so I have written it for the record for our configuration, maximum bandwidth is 48 PRBs, okay, right, and now how does it choose this CSRS, okay, is amply clear that if GNV wants to sound out the full bandwidth path, okay, it will choose CSRS, okay, this number will choose such that MSRS 0 is close to, is as close as possible to the complete bandwidth path, okay, right, in this case it could happen that 48 PRBs is the complete bandwidth path also, okay, and I am not saying that how much is the complete bandwidth path, okay, I am only saying that for the example I have taken, okay, it wants to sound out 48 PRBs, okay, now if you look at it, why it is choosing CSRS as 13, there could be 2 reasons, as I said, okay, that a complete bandwidth path itself is 48 PRBs, that is one reason, fine, okay, even if the bandwidth path is large, it is interested in only finding out the channel in only 48 PRBs, right, okay, so both these things would happen, we do not know what is going on at the background, but that is how it is configured, fine, okay, right, and so then what happens, okay, if only partial bandwidth path needs to be sounded out, then GNV chooses CSRS that MSRS0 is less than bandwidth path size, okay, so what is happening here that if so, that is what I said, the second case, bandwidth path is larger, but it is interested only in finding out the channel in 48 PRBs, that is what is being said here, okay, even though the bandwidth path is larger, okay, you base station is finding out, interested in finding out the channel only in a smaller part of the bandwidth, okay, so maybe it has picked CSRS accordingly, okay, right, and so what will happen, if so, if CSRS has to be picked accordingly, whether it is like the complete bandwidth path or only a part of the bandwidth path, okay, right, that is what is being said, fine, so this is clear, okay, right, before we understand all of this, okay, I will have to give you some little more information, non-technical in terms of standard, but technical in terms of theory, okay, right, so this notion is clear that how SRS will be configured, okay, right, why all these options are there, we will get back to it, but we will understand this, before, this is fundamentals, now all these parameters, they help in something called hopping, okay, hopping is basically you change the frequency and then transmit the SRS, we will come to it, okay, but I am just telling that what is hopping, you keep hopping from one frequency to other, just understand that, okay, and we will see what, how that hopping is happening, okay, now SRS is basically transmitted in uplink, correct, okay, and you see that your UE has a limited battery, there is a limited power, okay, right, and typically UE transmits at 23 dBm of power, which is 250 milliwatts, okay, right, and if a UE has 100 megahertz of bandwidth, okay, bandwidth part is like 100 megahertz, okay, now if the UE has to send the SRS in complete 100 megahertz bandwidth, okay, right, so the energy per SRS will be very low, because you are dividing that 250 milliwatt of power in like 275 or 272 PRPs, right, so per PRP power will be very low, correct, okay, and each OFDM symbol, correct, okay, now so if the power transmit power is low, channel estimates will be bad, correct, if the UE is far from the base station, right, okay, so base station knows, okay, this UE is far from the base station, okay, and then what it will do smartly is and that is how the standards are designed, okay, that it will tell the UE to transmit the SRS in two parts, okay, right, so what is happening, okay, so UE has a limited transmit power, transmitting SRS over a large bandwidth part will reduce SRS power density, power density is power per unit bandwidth power, okay, fine, so what it is saying that is what I said, per RE the power will come down, okay, in each OFDM symbol, what happens here is, what is the solution? Solution is to configure the UE to transmit the SRS over a small bandwidth and then hop, if this is the complete bandwidth part, okay, what are we saying here? That how are we configuring the SRS? How will the base station configure the SRS? It will say in this symbol transmit the SRS in this part of the bandwidth part and then hop, hop means change the frequency and then transmit in this part of the bandwidth, correct, so this happens in one OFDM symbol, this happens in other OFDM symbol and why the hopping is done? Because if SRS is transmitted in this, complete SRS is transmitted in this OFDM symbol, energy will be low, okay, but now what is happening here is part here, okay, right, so half is transmitted here and half is transmitted here, okay, so 250 milliwatt can be used for this part, in the next OFDM symbol you again have 250 milliwatt of power, okay, so that 250 milliwatt can be used here, so the power is double, right, okay, so that is why this hopping is happening, you see that hopping is because you are not transmitting over the same frequency, you are hopping the frequency, correct, okay. And we are forced to do that because transmit power is low and if we configure it like this, so the full transmit power can be used for these many PRPs, then again the same full transmit power can be used for these PRPs, okay, so it helps in improving the link budget, right, okay, so this is an example of intra-slot hopping, you are hopping within a slot, inter-slot hopping will come to that later, okay, right, so what are we saying here? So, the SRS power density is increased by 3 dB compared to sounding the full SRS bandwidth in a single slot, okay, what are we saying here? That here if you transmit the complete SRS in this symbol, right and we transmit it like this, this will have a 3 dB higher link budget because here will be spread over the complete bandwidth, here it is only the power is spread only over the half of the bandwidth, okay, right, basically then your effective power is 250 milliwatt, but it is only being used for fewer number of PRPs, right, so the effective power is doubling, same is the case here, okay, so that is what is said over here, right, okay. So, and this hopping will basically will discuss in more detail, okay, I have more slides, but I think the time is over, so I will continue discussing that in the next class also, okay, thank you.