 And yeah, the next talk is going to be not really about OSMO-COM. It's rather a niche or related topic, which is really about a bit of radio frequency planning. I'm absolutely not the expert in this. I don't really feel very qualified to talk about it, but still I think for the fundamental concepts, my knowledge is sufficient. Is anyone in this room who is actually doing actual frequency planning for radio networks? You've done it, okay. Because what we see quite a bit in terms of the user base is that, okay, there are some people who manage to download or install or compile or somehow configure and use the OSMO-COM GSM stack or GPRS or whatever it is. But when it comes to actually deploying the network itself, there's a lot of, I mean, these people, they might have Linux's admin skills and so on and so on, but there's often a lack of understanding of some basic radio concepts in order to install that. And interestingly also, it's not, apparently not always easy to find somebody who can help them there, or at least they think so. So I thought I'd introduce some basic concepts. So of course, in radio planning, there are many, many different parts of radio planning. I mean, first of all, there is the question about the path loss and link budget in which I'm focusing here, which basically is making sure that the signal really reaches where it's supposed to reach or evaluating what kind of equipment do I need or to reach a certain destination or the other way around what kind of equipment do I need to reach the destination or how far can I go with my signal. Then of course, there's all kinds of other aspects in real network. There's issues of interference management. There is issues of neighbor cells interfering with other cells. So you have a downtilt of antennas and so on, so you can go on forever. But as I said, I'd really like to look a bit at the basic part of how you handle this. There's going to be a lot of talk about DBMs and DBI's and what's and so on. So these are basically the units of measurements that we use. And what we focus on right now in this talk is something called path loss. And path loss is basically, as the name sort of implies, the amount of loss that you signal encounters along the path between point A and B somewhere. So your point A might be at your base transceiver station output. So at the coaxial connector back here is, let's say, point A. And then there is maybe some cabling. There is an antenna. There is the actual radio link. There is another antenna that receives it, which is built into this phone, for example. There's no cabling inside the phone. But then anyway, the phone at some point receives the signal. And the overall loss between the transmitter output and the receiver input is the path loss that we encounter. And GSM operates in frequency duplex. So the uplink and downlink frequencies operate in different parts of the frequency spectrum. They also behave slightly different because even though it's only several tens of megahertz, already the propagation can be different. Between those two, so actually you have uplink path loss, which is the path loss from your phone to the network, and you have downlink path loss from the network down to the phone. And they also need to be considered separately because the sensitivity of your receiver in the phone is different than the sensitivity of your receiver in a base station. What does sensitivity mean? Sensitivity means the amount of signal required in the receiver to still be able to decode the signal properly. So typically a base station, since it's a professional grade equipment, has a much higher sensitivity than the phone, which is a consumer grade equipment manufactured at much lower cost in lower power margins in terms of electrical power consumption and so on. So you have to look at that separately because the different sensitivity and also the transmit power of the device is different in the both directions. So let's see. So the most simple path loss to look at is the so-called free space path loss, and that basically is path loss in outer space in a vacuum, where there's nothing in between, there's no particles, there's no water molecules that sort of attenuate the signal and so on. So there's basically nothing in between, no buildings, no nothing. That's free space path loss. So free space path loss is relatively easy to compute, but it's also not very useful because unless you do, I don't know, I'm not sure whether it's useful if you do communication with the Voyager probe or something like that, but at least in terms of GSM communication with OsmoCom, I don't think free space path loss is terribly important. So what do we do? We need to somehow estimate the path loss in a given real-world situation and as many different approaches how you can do this, if you go to operators today that do real network planning, they, of course, they have this specialized industry that produces software tools for the coverage analysis and they use really high-detailed 3D geometric data of the buildings and of the terrain of the land and then even maybe today using ray tracing methodologies or basically they actually like ray tracing, you know that maybe from rendering. It used to be popular some decades ago as a method of rendering graphics and basically you are tracing the path of a simulated light source and I mean a radio source is nothing else. It's a source of electromagnetic waves. So you basically try to compute as many waves possible and see where they get reflected and where they get attenuated and so on and then you have some realistic estimates. But most people who build networks with OsmoCom don't have access to such technologies. So basically we have to go back some decades and use the more simple rules of thumb that are available and I'm getting to this very shortly. So, but generally what kind of things impact our path loss? Well, there is the height of the antennas both on the receiver and the transmitter side. There is whether it's line of sight or non-line of sight operation. Line of sight of course is much easier than non-line of sight like here. Reception is very bad in the basement of course. There's no line of sight to any base station. Well, this one, yeah, but not out there. So that's why you should install Osmo BTS based devices everywhere. The geography terrain of the region, do you have hills? Do you have mountains out of which material maybe even to some extent? Also the vegetation, particularly the foliage, the leaves because they contain water. Even the attenuation is different if you're in a forest or you have signal traverses of forest, whether it's winter or summer at least in non-equatorial regions where you actually have seasons and in winter there are no leaves on the tree. The attenuation is different than in spring or summer when you have lots of leaves. Then of course any type of construction and the type of materials used in that construction, height of buildings, distance, the frequency band most importantly which is why frequency bands in lower spectrum like 850 MHz, 900 MHz, GSM are much more valuable than the 1800 or 19 MHz bands because the signal propagation in the lower frequencies is of course much better. All these factors determine the actual attenuation of the radio wave but then we also have another aspect which I'm not going to go into a lot of detail here which is called the multi-path fading which happens because there's multiple paths that a wave travels between sender and receiver so you have reflections like let's say between the room and here there's a direct signal propagation there's one that bounces on the left wall then goes to the right and goes back there and so on. Basically the receiver doesn't receive one wave but it receives an infinite number of waves over an infinite number of propagation paths that get reflected and the reflections mean that the distance is longer and therefore there is a time shift between the individual wave fronts when they are received in the receiver and that really is causing an effect called multi-path fading and typically in those simple models you just add a couple of dB more margin in your equation to cover for that loss that additional loss that the multi-path propagation does. So over the decades people have done all kinds of models as I said there is expensive proprietary tools but I mean we have some general rules of thumb that we can use as some models let me actually make that slightly smaller so we can fit it on one slide hopefully like this. So there's a couple of different models that the ITU developed for radio planning specifically in the GSM bands. This was the early days of GSM when they didn't have the 3D ray tracing tools and there is basically a terrain model there is a one woodland terminal model I like that word so the idea is that basically one terminal normally the phone is inside a forest and the antenna is outside of the forest either by height or because it's actually literally outside the area covered by the forest so you have the base station antenna in the open space and the terminal inside the forest and that's the one woodland terminal model. So and you have these by types so you have a terrain model that just tries to cover the terrain, the geography you have the rural foliage model which tries to estimate the losses you have in terms of trees and vegetation and you have multiple city models divided in urban, sub-urban, open and so on and not all of the models exist for all the frequency bands so you have to be careful there you cannot use all of the models in all frequency bands and then you have in the city you see you have all these Okumura Hata and other modified models such as the cost 231 Hata model and then there's also a Valfish Ikegami model they have always named after the inventors of those models people who came up with and they're actually very simple how do you come up with such models? Well basically you take lots of measurements and then you try to find an equation find a relationship a mathematical relation and put a formula that sort of reflects the observations that you have seen so this is not a deterministic model but it's just based on observation and trying to model the real world observations so there's some references about the path loss models you can find it in the slides and also there is a chapter in the Osmo GSM manuals which contains all this and more text associated with it I put it in there so I think it's not really rendered into any of the manuals yet but all the text is there and I'm still wondering where to actually put it or whether it's yet another manual okay so if you look at the RF power in the wires link we have the signal strength in dBm that's decibels for milliwatt in the transmit side here on the y-axis and on the x-axis we have the distance so our transmitter transmits some signal into a cable that's the cable that feeds your antenna and the signal sort of effectively gets weaker as it passes along the cable because cable has a lot of loss then you have an antenna and the antenna has a gain so basically instantaneously in this model your signal gets much stronger and then it gets emitted with something called the EIRP the effective isotropic radiated power and that's also again a hypothetical idea of how much better your signal is than the idealized isotropic radiator that doesn't exist in reality and then over the actual air interface you have the loss of the signal along this curve and then again you have a receiver antenna which has some gain so again instantaneously your signal gets better you again have cable losses well in most mobile phones you don't but just for the sake of illustration let's assume you have a car mounted phone with a cable between the antenna and the phone and then you at some point you have a certain received signal length signal power and this year the dashed line is your sensitivity of the receiver so as long as your signal is on average significantly higher than the receiver sensitivity your signal propagation is successful as soon as you go below this Rx sensitivity you have signal loss at that point and in order to ensure that it stays mostly above the sensitivity you have something called a margin you add that margin when you design the system in a way that you say well I make sure that my signal is always stronger than that margin so whatever may happen maybe there is rain or some additional factor that causes more attenuation or the multipath fading so I have some reserve in my budget which brings us to the link budget and the link budget now uses the path loss to compute or to basically see what put in all the factors and see what do we need actually so we have these antenna gains we also have other parts like duplexes splitters maybe some additional power amplifiers low noise amplifiers and so on and we put this together and the simplified link budget equation in the end is basically the received power equals the transmitted power plus all the gains that we have in terms of antenna gain or amplifier gain so we are expected by all the losses that we have in terms of cable losses or propagation losses over the radio interface and if we add all this up put this into illustration here so let's say we have a base station that transmits at plus 20 dBm we have a bit of cable that has 2 dB losses we have an antenna that has 10 dBi gain 14 dB attenuation over that path whatever we will get to computing this also that's basically the result of your path model so you use the model like your urban or sub urban or whatever model and it will tell you over 5 km you have such and such amount of dB you put that in here and then again you have some antenna gain cable loss and then you have the actual power assuming that let's say we have a receiver quite good receiver like you find it in good base stations of minus 109 dBm sensitivity here and we have a phone on this side that transmits at 20 dBm like if we actually try to resolve this so we have plus 20 minus 2 so we are at 18 then we have 10 we are 28 then we subtract the minus 114 what is it? 86? we have another 10 dB here so we are at 76 so minus 86 minus 2 so at minus 74 anyway we are much higher than our received sensitivity here so in this particular example it would work my math might be wrong but in any ways I have a numeric spreadsheet for that actually I wrote a numeric plugin for that so we will get to that so what do we have here still? this slide so if we look at the uplink we have the mobile station, the antenna, the path, the BTS antenna this assumes that we have a duplex head, the receiver and at some point we have a BTS that is outside of the slide but I can assure you it is a BTS now how much power does the phone transmit? the typical transmit power level of all the phones from the late 90s till today in GSM looks like this if you go for really early phones you can still find 8W capable phones but these are not hand held units this is a suit case type of GSM phones the portable phones you can still find with higher transmit power but this is basically it so on 900MHz for 850MHz most of the phones do 33 dBm which is 2 watts, some only do 30 and on the oh sorry this is the other band on the 1890MHz most of the phones only do 1W and on Edge it's even much less so this is the E2 power class for the low band and for the high band has even much lower transmit power so this basically explains why the coverage radius for Edge is smaller than the coverage radius for GPRS not only is there a more complex modulation where you have higher probability of some bits being detected wrong but also you have much lower transmit power to begin with so your signal is much weaker so that's why the most robust communication is in the low frequency bands so you have better signal propagation and stronger signals by phones that can transmit more power and if you look at the more modern standards like UMTS and LTE it's much less so the transmit power of the devices becomes less and less as the radio standards evolve to some extent it's because the receivers are getting better but to some other extent it also means you just need many more cells to cover your area than in original GSM networks where you had really really large cells they also had very aggressive the newer standards also have very very aggressive changes in the energy to bit ratio and that helps yeah it's correct okay in the downlink it's pretty much the same we have the BTS and the BTS transmit power of course depends on the BTS model and hardware that I use and maybe I use an external power amplifier and so on and so on I have a fix me on my slide because there was a fix me in the original document that I made the slides from most likely so in this case in this direction what I spoke about is the transmit power on the mobile station but what scrolled out of this due to the zoom is that what's the sensitivity of my mobile phone in the other direction of the BTS in the other direction this is mobile station to BTS so the spec for 3GPP specification for GSM says it must be at least minus 104 dBm on the BTS side in reality it's more 108, 109 maybe even a bit more that sort of what you get from modern BTS equipment on the other end and in the reverse direction when the BTS transmits you check the beta sheet of your BTS and the receiver on the mobile station must be according to the spec must be at least minus 102 dBm older phones are also in that direction but modern phones actually are much better so sometimes you see phones that are as good as base stations now in terms of received sensitivity because of the advances in technology in general but if you want to be on the safe side assume that minus 102 dBm is the sensitivity of the phone so yeah that's the minimum yeah it's the minimum sensitivity a sense that a phone can always have better sensitivity so go down to minus 108 dBm or something like that but it is a phone that would only have minus 100 dBm would not be 3VV compliant they probably still exist of course but it's not compliant at least at least for example in Europe you wouldn't be able to declare it CE compliant if you cared about the regulation so now what can we do in terms of optimizing the link budget that we have so the link budget basically is the difference between what we transmit and what we receive and the margin we have in there and if we determine that well we want to we want to have a coverage area of let's say 5 km from there to here and we run all those figures we use one of these models which I will get back to let's make a spreadsheet or something and then we see the signal arriving at the phone is minus 120 but the phone has a sensitivity of minus 102 so there's like 18 dB difference so it's not possible to make it so what kind of options do we have well in the downlink we can always add a bigger power amplifier if we have the electrical power available and the cost of the power amplifier and the radiated thermal energy and so on it's not a problem we can increase the power amplifier and if you look at classic old BTS hardware you can find BTS's that have 40 or 60 watts of transmit power which is like okay did they want to roast chicken so but yes you can do that there are reasons for that because they have long cables and basically two thirds are lost in the cable before it hits the antenna and so on but anyway you can have bigger power transmit but the problem is of course the uplink direction in the reverse I mean you don't want to carry around a large power amplifier with your phone and some batteries and that's not realistic and brain cancer so so yeah you can of course try to reduce cable losses this is a very important factor that many people who have not really looked at this are not aware of is how much you lost in cabling so particularly in classic BTS setups where you basically have a high tower you know very high tower and all the BTS equipment is basically on ground level and then you have lots and lots and lots of cables like 50 meters cable or something like that and then of course over those 50 meter cables you lose pretty much all of your signal and that's not very useful so always try to get your BTS as close as possible to the antenna because then you avoid these losses in the signal increasing the height of the antenna also always almost always helps I mean certainly not beyond a certain point but that's one of the factors in all the signal propagation models is the height of the transmitter antenna and the higher you get the more likely it is irrespective of terrain or buildings that you can get closer to or into a line of sight situation as opposed to a non-line of sight situation okay yeah there's some other stuff that I'm not going to go through this is just also from this paper about like coaxial cables and coaxial connectors and what you should do and you shouldn't do and what duplexes are and how you can power amplifiers in the link and so on and so on going to skip all of that and rather I'm going to just quickly open this spreadsheet that I made just two two aspects of that the path loss models that we had originally the different ones that the hilly terrain the one terminal woodland model and all these funny models it's not very easy at least I don't know how to put this in a normal spreadsheet just by entering formula I don't know I don't think it's possible at least not with the kind of spreadsheet programs I am familiar with so what I did I wrote a numeric plugin which you can basically do in python so you don't need to compile and link it against the specific numeric version that you have so numeric is a spreadsheet to begin with program and so there's a numeric plugin that I have published in a git repository where which you can load in your numeric and then I have a spreadsheet that uses that numeric module to basically execute all these different models to estimate the signal propagation loss and then in there I also have the different like different antenna types and so on so let's just have a look at that very quickly we'll need a minute or so to open it so this is too small to read most likely now I have to figure out how to zoom in numeric I don't think I did that really much before let's try with 200% 200% actually let me change that slightly like this so basically there is a start slide where we say what's the transmitter power in downlink and uplink the frequency and so on and it computes various other values so let's say the uplink here now for 2 watts which equals 33dbm so that's already computed from the 2 watt that you input there this is the actual transmit frequency on uplink then we assume we have about 2db body loss that's your human body because you have a hand and a head and so on next to the antenna we have no duplexer inside the phone I mean none that is added external to the phone let's say we don't have cable and so on so we end up transmitting about 31dbm on the receiver side let's say we have no antenna gain minus 108db sensitivity and so on so we end up with a total system game of 138db and then from that system gain we reduce some factors like the fading loss the indoor penetration well indoor was not used here and then we have some margin we subtract and we say well the total path loss is 100 or the total link budget is 130.5db in this example now this is all the details but if you use that to compute this let's say for a this is for a 2 watt base station again apparently I have to zoom for each sheet separately so this is a 2 watt BTS assuming a 25 meter antenna height and then it computes all the basically we have different columns uplink downlink for no antenna gain a 2db antenna gain 5db antenna gain 7 9db 11db antenna gain and if we look at the estimated coverage that it computes the font is too large that's why the first line well this is the free space path loss which is not interesting anyway so the free space path loss would say well you know you get 38 kilometers in downlink in high-end kilometers in uplink and it's like yeah okay but then if we look at these the different propagation models then we get more realistic figures here right so this is without any gain at the antenna we can see that the the sorry what was that the downlink is first so we can see that the downlink in a large city would be 750 meters and the uplink would be 1.2 kilometers so we can see that the link is not but is not balanced in in in downlink we only get 750 meter but in uplink we get 1.2 kilometers which means we are constrained by the downlink so we should add more power or more gain to the downlink so we balance the link that because we need both way communication and this propagates down here so even if we use a relatively good antenna with with 11db gain we still have 1.5 kilometer in downlink so it's still not balanced we need more power on the transmit side to get to the maximum range in in a large city so if we go for the sorry for the one terminal woodland I think this is the second line here so 2.9 versus 4.5 kilometers with one terminal in the woodland and if we go to a 10 watt bts we will see with these figures that I put in there of course it will not apply to all but still it should give you some educated guess where was the zoom so if we look at this for 10 watts the same computation still 25 meter height of the antenna mast on the bts 1.5 meter height of the phone like sort of the level of the phone when you make a call above ground and if we again look at the 11 dbi antenna here at the end then we see it's almost equal according to theory so in that case we see that about 10 watt seems to be a good transmit power to balance the link assuming a receiver on the phone hand side that only performs as good as the spec requires it to right this is the minus 102 ebm and basically the all this difference that we see here that we because the phone still transmits the 2 watt but the bts transmit the 10 watt that's only due to the fact that we assume the bts receiver is so much better than the the phone receiver or to a large extent at least and if the phone is much better than this of course doesn't work anymore but you don't know how old or how good the phone is so if you want to be on the safe side you can make the only the assumption that it's as good as the spec requires it to be okay so that's basically some short intro into this to to get some some educated guess I mean this is not more than educated guess you can get from these kind of simple models but at least it is something to start with if you if you want to deploy a base station somewhere questions yeah can you find the values for the phones something like under fcc where they need to publish the tests it's a good question I don't think I have looked for this in terms of specific phones on and I don't really think it's something well is it something that needs to be tested as part of the fcc approval I don't even know that whether it's mandatory for a phone to to do the receive sensitivity testing I cannot really tell it would be interesting I could think it is the case but I don't know check the fcc database and see if you see receive sensitivity indication for phones normally I mean most consumer phones you don't get real data sheets that really sucks but if you look at data modules like gsm modems actually from a lot of manufacturers you can get the data sheets and they have I've seen from several manufacturers actual sensitivity figures on there so there it's more likely that you get some hard data on the real technical parameters as opposed to consumer phones okay yeah we need the mic but it's still off I just wanted to know if there's kind of like an error all the values are kind of very absolute seems very precise but the conditions are kind of static I just want to know if kind of like if it's raining some storm if kind of this error if you have kind of a percentage how the value can change well you put always some db of margin for these kind of things additional margin normally so let's say for the here on the on the first slide I had basically all the parameters are described so for shadow fading we have here a margin of 7.5 db and that's computed again by some other values and if you want for indoor penetration I think you put another 3db or something like that I'm not the expert on this but you can check and there are some books on radio planning and so on and if you want to do these kind of this is all full of errors everywhere as I said it's an estimated guess but I mean there are also some like standard values to put in there but rain is not that big an issue I mean we're not talking about the massive amount of difference here not at the frequencies that we are transmitting because it's more of an issue with Wi-Fi for example because 2.4 GHz is the resonant frequency of clusters of water molecules or something like that and which is why you use it in a microwave often right but the frequencies don't water for example is not particular it's not a resonant frequency or something like that 3db I just a quick comment because this is great it's really important stuff I think and not just with cellular but also FM broadcast or whatever during years without having any specialized test equipment wandering around like trying to figure things out in a very non-scientific way so just a quick comment due to the work again of the osmo community the RTL-SDR you can plug that into an android device and basically for quite cheaply get yourself a little spectrum analyzer that will give you figures which is a lot better than maybe wandering around and your coverage area counting bars on the telephone and saying ok I have signal here I don't have signal there this thing will actually give you some numbers if you can standardize your antenna in some way so that you have a reference point then you can map if you have an installation you can actually map it to some extent and see what you're actually getting in the real world in your installation area I'm not sure whether RTL-SDR is really such an improvement there sure the signal bars is not that strong but if you have a monitor mode that gives you the dbm reading or on android there are apps which give you the raw value of course a phone is not a measurement device so the tolerances that are permitted by the specification for these kind of indications I think it's also plus minus 3 dB so it's massive compared to an actual measurement device but still I think it's not really bad to use phones for doing coverage testing the bars is a very rough granularity but if you can get to the actual receive level I think it's still good sorry you don't have that mic I just had a general question about the F2P requirement sorry I think it's not on the LED should be off the top LED, the mute okay got it positioning of the FCCs that receive sensitivity or performance in general is not tested so that goes beyond that that also covers interference from adjacent bands is one example where they also do not test that at all either so came up earlier I suppose this is for the phone side for the base station it's probably part of the testing or also not I do not know about the I guess it was probably the same case actually where that would fall under the manufacturer's test for their performance reasons in Europe the receiver performance is part of the CE but the filings are not public so basically you need to as a manufacturer you need to actually you don't need to file it in the first place you self certify and you have all the documents and if a regulator ever inquires then you have to provide the documents but the receiver performance is a key part of the regulation so because they basically want to achieve efficient use of the spectrum that's the regulatory goal and it's just mandating let's say it out of band or something like that is insufficient and that's why there's the strong receiver requirements but yeah no public information I can provide one comment on the handset side the reasoning there is to say the manufacturer can make a phone as cheaply and as performing as possible and if the user wants to buy it then they should be allowed to buy that product the FCC testing assumes the market has the utmost incentive to like pursue performance however it works so you're right on that the only numbers that they care about have to do with RF exposure which is kind of related to all this but also quite separate that's the stuff that gets reported on I think okay yeah then we're already behind schedule again we have to switch to the next topic which is the afternoon break