 to messages from Pluto, signals from Space 101. You know how they say that there is absolute silence in space, but for anyone listening on the right frequencies, there are still plenty of signals to pick up. And our next speaker, Inko, is a PhD student in Stuttgart, and he's been working with satellites at the university, and he's going to tell us all about it. So I would like to welcome you, Inko, and I would like to ask you to welcome him with a warm round of applause and have a lot of fun with the talk. Thank you. So when Sputnik was launched and started transmitting its signal, most of the people I guess in this room have heard the characteristic beep. But what was this beep actually? So Sputnik actually had two different transmitters inside of it. That's why it has four antennas, two for each transmitter. And they were alternating an activity so you could hear a beep either on the one channel or on the other. And it was not just a beep, it was not just a signal, yeah, we are here, we are up here, we can reach you. It was actually encoding data because Sputnik was a sphere, so you could more easily put it under pressure because electronics and vacuum was still a difficult topic back then. So inside the beep, by changing the length of the beeps, so either using the one frequency more or the other, they encoded if the pressure dropped under 0.3 per five bars or if the temperature went outside the range where they were comfortable with. So let's take a step back and look at the problem more generally. How can you encode data onto a radio frequency wave? If you just send a sine wave, you basically are sending no information. So you have to make changes to it. So one kind of changes you can make to a sine wave is changing its amplitude. So if you want to have a different symbol, you use a different amplitude. Another way to encode data is to change the phase of the signal. So if you look here in the diagram, you can see that on the boundary of the symbol, the phase jumps a certain amount and then jumps back. So we have a change here and we can encode data with it. But of course you can also combine both techniques to do amplitude and phase modulation at the same time. And modulation is the name of this technique to imprint information onto the RF carrier. So now if we want to have a more general approach and see how can we describe these kinds of modulations, usually what we use is a constellation diagram. So in a constellation diagram, you can see on a 2D plane usually points and these points encode the information of amplitude via their distance from the origin and phase via the angle to their vector from the x-axis. And now we can just define our valid symbols inside this constellation diagram and assign bit values to them. So for example, here you can see we have four valid points. So we have four different symbols and we can assign two bits to each of them. And now if we want to transmit a series of bits, we just use different valid points from our constellation diagram. So as you have probably already guessed, so if we want to transmit more information, more bits with the same symbol, we just have to use a constellation diagram which has more valid points. So for example, here 32 points and 32 APSK with five bits per symbol. So why don't we just use infinite points in the constellation diagram? Here we have to take a look at what happens when we actually transmit the information. So when you do the transmission, we can pretty accurately hit these valid points in our constellation diagram. But then when we receive them, we see that they are not actually there where we were expecting them. So they're not always hit the same point. So at some conditions, we can't validly identify the correct symbol anymore. And the effect we are seeing here is noise. So noise is changing up the point in the constellation. And at some point, we can't identify them anymore. So what else can we do to increase our information which we transmit? As I already said, if you just send a sine wave, basically you are not transmitting any information. And if you look at the spectrum of your transmission, you are just seeing a very, very thin line, basically infinitism is small. But now, when we start making changes doing modulation, transmitting information, we are seeing the spectrum widening up. And actually, the more symbols we transmit, the more frequently we change our symbol, the wider the frequency plot gets. So we are using more bandwidth. And that also means other people can use the same bandwidth. So we have a limited resource here. So we can't also not just go infinite with our symbol rate. Another problem we have here is that if we want to accurately demodulate so we cover the information, identify the IQ constellation points, we have to take into account this whole frequency range. And normally what you have is in, let's say, natural noise situations that you have a noise floor which is approximately equal across your band. So you just have over the frequency a certain noise band. And the more bandwidth you use into your application, the more you have to also take in the noise within the band. So that means if you have a high symbol rate, you're changing lots of symbols to have a high bandwidth. So you're automatically also getting more noise power into your system. So here from the left side where you have the symbols and the noise floor on the right side, you can see what you can get if you do a band pass filtering. And you can obviously see you get more noise power if you have a wider signal. So how to look at this, how to compare different constellation diagrams. Usually you compare the bit error rate here on the vertical axis. So the probability that a certain bit is validly transmitted over something that is called the energy per bit over the noise density. So what does this mean? As I said, you can vary the symbol rate, you get more noise, but you also get more information. You can have a different amount of spectral efficiency, so the bits per symbol. And to get a more general view of it, you just compare how many energy you can put into each bit. So usually you get a curve like here. So the bit error rate drops. The more energy you put into your signal, so the more energy you have per bit. And the more complex your constellation is, the closer the points in the constellation get to each other, the higher the bit error rate is for a certain EBN0, so for a certain amount of energy you put into the bit. Because they're closer together, they are more easily disturbed. So what can you do to increase this energy per bit over the noise floor? So this is actually at the receiver. So we can just try to focus the signal more into direction. We want to receive the signal. So for a receiver which is sufficiently far away from the transmitter, it actually doesn't matter if the transmitter transmits all the energy towards the transceiver or a higher energy but all around it. So what we are typically looking at is the equivalent isotropic-gradated power. So the power we are seeing, if we are assuming that the transmitter is transmitting into every direction, so omnidirectional. And this is basically just the amount of power we put into the antenna times the gain of the antenna. So if we have an antenna which has a very, very focused beam, so we have lots of energy towards our receiver, we don't need that much input power to get the same result at the receiver. But there is a conundrum we have here. If we increase the gain, so the amount of power directed to our receiver, we also have to take into account that our beam width, so the amount of focusing is increased. That means if let's say we have an attitude control system or a satellite, we try to find our ground station, we have to try harder. Because if we just point the satellite a little bit away from our receiver, suddenly we have a huge drop in received signal strength. And so we can't really get the same energy which we had before. So this is a general thing you have to take into account. If you have a more focused beam, you have a smaller beam width but a higher gain, so more energy on the receiving side and vice versa. But also if you increase the frequency for let's say the same general architecture of your dish, you also increase the focusing of your antenna. So if you have, let's say, 10 gigahertz instead of 2 gigahertz, you have to try harder to point your antenna. So what kinds of antennas we typically have on satellite systems? One kind of antenna is the quasi-omni-directional antenna system, which is used for, let's say, sending commands to the satellite when it's close to Earth or so. And also when there is an anomaly on board, the satellite usually tries to gain as much solar power as it can, so it points towards the sun with its solar arrays, and it might not really point to an Earth station anymore. So in this case, we still want to be able to transmit data to the satellite telecommands or receive information about the satellite's status. So we have to have an antenna system, which is approximately equally good at transmitting into every direction. So for higher frequencies, you typically need more than one antenna to get this quasi-omni-directional characteristics. So we typically have two inter-opposite directions. Then we have most often payload or communication data streams, which often use high-gain antenna, so they have a more focused beam. Get a higher EBN0 can transmit more bits without getting into trouble with the noise floor, but have to point their antenna. So here we have the normal operation, so we assume that the satellite is able to point its antenna, and we can use the benefits of having a high gain for the same power, so having a higher received strength. And then there are some, let's say, more interesting antenna patterns. Let's say you have a navigational satellite, GPS Galileo, something like that, and you basically want to illuminate a whole side of the Earth, but some parts, because as we all know, the Earth is round. So some parts of the Earth is farther away than you, but then the points directly beneath you if you point towards the Earth's center. So you basically would need a little bit more gain towards the far sides of the Earth, so the Earth limp. And so you here have antenna patterns, which are called esoflux patterns, which don't actually have the highest gain in the bore side, so it's directly vertical from the antenna surface, but more to the sides. So it has a little dim in the middle. But this is good, because now you have about the same received strength on the whole planetary surface with the surface. But yeah, that's one side, so we now have optimized our transmission. Now the signal actually has to get to our ground station. And there are lots of things which can attenuate your signal. First of all, and usually most of all, it's just the distance. So if you are transmitting radiation, electromagnetic radiation, the power drops off with the square of the distance, because you just spread the energy over a larger and larger surface, which goes up with the square of the distance. So you get much, much lower received strength if you move farther away, or the satellite is farther away. But then on top of that, you get other effects. For example, if you have a relatively low frequency, let's say sub 1 gigahertz, ionospheric attenuation can become significant, so you get a drop in the signal or even reflection if you just hit the upper parts of the atmosphere, which are ionized. So yeah, charged, basically charged particles. And then everything in the atmosphere. So rain is usually a huge problem. If you have a high frequency, sometimes maybe you experience it if you have a small set TV dish. Then if you have rain or snow or so, you get a significant drop, maybe even so far that you get signal loss. But yeah, in general, every water vapor in the atmosphere is a problem. And rain is an extreme case here. If you, let's say, are not that good with selecting your frequency, you could also hit just absorption spike within the gases in the atmosphere. So for example, oxygen has an absorption spike in the tens of gigahertz range, where you just really can't get much signal through because just the oxygen in the air is attenuating your signal. So now we basically know how our signal gets there. What is with the noise? So how is the noise for determined? Much of the noise we see in satellite systems are actually generated also outside of the receiver itself. So as a rule of thumb, you can say that everything that absorbs energy at a certain frequency also emits very well at this frequency. So again, we have our old enemy rain here, which if you have a rain shower and you try to receive through it, you suddenly get a much higher noise floor, which adds insult to injury regarding the attenuation before. Another major source of noise in our, let's say, solar system is the sun. So the sun is not only powerful, as we all had to learn in the last couple of days in the visible spectrum, but also in the radio spectrum. So if your satellite is passing in front of the sun and you would have to point your antenna towards it, suddenly you get a lot, a lot of signal which can swamp out your signal from the satellite and you can't track it anymore. So this is an actual problem with satellite systems. But then if you get better, you have very good receiving technology. At some point, you hit, let's say, upper limit because, as you know, there is a cosmic microwave background, which is behaving like a black body, radiating about 2 to 3 Kelvin. And this is also a certain noise floor, which you basically can't get under because you will always see it no matter where you look. And then for certain antennas, if you try to be very efficient, you of course try to use the whole dish you have paid for. Then it can be that the dish has certain side lobes and also back lobes. So it's not only that the dish has the main beam focused on the satellite, but also some other direction you might get reflected energy into your dish. And then the earth surface becomes significant. So if you have a dish which tries to use much of its area, you often have bigger side lobes. And you get a problem if you're pointing to low elevations because suddenly the earth surface radiates into your antenna and increases your noise floor. So maybe we can just take a look in this example satellite. So what I'm showing here is the uplink path. So let's say from ground to satellite, maybe to transmit telecommands. So what has the satellite to do or things like that? And let's say, let's see how the signal is received. So at first, on the input antenna, we have a bandpass filter, which is relatively coarse. It's basically just there to split apart the parts of the frequency we use to receive data from the ground, from the parts we use to transmit to the ground. Because otherwise, as soon as you would switch on your transmitter, you would just swamp your receiver. After that, we have here a low noise amplifier, which is used to increase the power we have on your incoming signal. It has to also travel through all the attenuation, the free space. So it's relatively low, and you have to boost it. So it can use it in the further stages when you make changes for it. And the first change you are doing to it is, this year, this is a mixer. Just in case somebody hasn't dabbled with RF electronics or so, what does a mixer do? Basically, you can input signals to a mixer often to, let's say, an information carrying signal and one, basically, a sine wave. And what comes out is the difference and the addition of the two frequencies. So typically, out of such a simple mixer, you get the carrier itself. OK, now here down, there should be the original baseband signal, and then plus and minus the frequency here of this information wave. And now, if you just need one of them, you can filter it and then have your original signal on a different frequency. So what you're doing here, basically, is just getting the 7, 8 gigahertz signal you got from the ground and mixing it down so you can more easily decode it. So afterwards, you have the span-press filter, again filtering out unneeded parts of your mixing process. And you have another down mixer. So you have a two-stage down convergence here. So now afterwards, you have changed the frequency or signal occupies from your very high couple of gigahertz range down to basically 0 hertz and a couple of megahertz, or even kilohertz. Then you have another gain stage before you go into your analog digital conversion and do more changes and detection in the digital domain. What is interesting here is that based here on your controller, it also determines how much energy it sees on its input and also within the small, small bandpass of your input filter here, where the carrier actually is. So where the signal is transmitting on. And you can use this to form a tracking loop and adjust the frequency it puts into the mixer here. So it can adjust for, let's say, wrong frequency on a receiver because it just tunes until it still sees in the middle of its receiving band. So this look in practice. So typically, you have a ground station which is trying to transmit on a certain frequency. But then, due to, for example, Doppler shift or an agent oscillator on your spacecraft or whatever, the spacecraft sees this signal on a different frequency. And now if it has just a small box where it can decode the signal, of course, this wouldn't work. So what the ground station does in the beginning of the process is it sweeps. So it just changes the signal around. And at some point, if it sweeps far enough, the signal is actually within the box of the input filter on the spacecraft. And then this tracking loop activates. And from there on, the spacecraft receiver always tries to keep the signal within the center of its box. So now, even if the sweep continues or stops, it has still tracked the signal. And can now, if the ground station starts to transmit information, it can extract the information always at the same frequency offset. So now, a different example for a different kind of satellite. We have a broadcasting satellite here. So what you would maybe receive if you have a satellite dish for TV at home. So typically, what we have here is, again, a band press filter for input-output splitting, then an LNA. So a low noise amplifier to just boost the signal a little bit so it can work with it. And then everything, the whole band with all the TV stations is mixed. So it's offset in frequency by a certain amount. Here, you don't have a closed loop tracking. It's just always the same amount that's offset. So now, the ground station would have to account for that if the oscillator ages or so and changes its frequency, it would have to be changed on the ground station side. What you then have is an input multiplexer, which basically splits the spectrum you get into it into certain transponder ranges. So a transponder is, let's say, a channel on your satellite. And within this channel, let's say, 36 megahertz was the standard, especially for satellites which were once designed to transmit analog TV. These portions are split within the input multiplexer. And each of them are fed through one of these amplifier elements with a band press filter afterwards, so they are independently boosted, amplified. And now, these amplifiers are power amplifiers. So what you get out of it is a very strong signal which is able to travel back the 36,000 kilometers to the Earth's surface and being received by your satellite antenna. Then we have the output multiplexer, which combines all these different transponder bands again to one single channel. You have the isolation via band press filter. Remember, now we have a different frequency because of the mixing here. So these frequencies here are different. They can't reenter your receiving side. And now, they're radiated out of your antenna dish or area, whatever. As you might have noticed here, we don't really have much digital electronic here. So the classical broadcast satellite is also called the band pipe satellite. It just takes everything that comes from the Earth, mixes it, amplifies it, and sends it back. It doesn't concern itself with what kind of signal is used. Let's say, if a TV station wants to change its video encoder from MPEG-2 to MPEG-4, it doesn't care. Even if it would just start to send radio or switch from analog to digital TV, it doesn't matter because the satellites are pretty dumb, at least the broadcast satellites. But this is an advantage because here you can make changes on the ground side. And the 15 years or so, the satellite is in orbit. It's still able to perform its function. On the other hand, remember the noise problem. The satellite here already sees on its antenna side noise coming from, for example, the Earth's surface atmosphere and so on. So the signal is already hampered by a certain noise floor when it's being received at the satellite. And then it is amplified. The noise is also amplified. Everything is sent back. So now you pick up additional noise on your way back to the Earth and, of course, on your Earth-based small antenna. So the system is not ideally ideal if you want to use a very low-noise system because it just amplifies also the noise it receives. The other thing would be a regenerative satellite, which actually tries to decode the data it sees, does error correction, remodulates the data, and sends it back. So I already mentioned forward error correction. So forward error correction we have to do to prevent that the noise we are getting is actually completely destroying a satellite. Because unlike, let's say, IPTCP, we can't usually ask for a retransmission. Of course, there are things where you can. But let's say for a TV satellite, you can't ask the satellite. Everyone can't ask the satellite to just retransmit the last frame. So you try to make it very robust, and that you do via forward error correction. So what kind of forward error correction could you do? For example, one type of forward error correction, you just take your bit stream or your data stream, you chop it up in blocks. Each block gets additional data. So we add some redundancy to the signal, and then it's put together again and transmitted. And another thing you could do would just take the input data stream and produce a completely different output data stream. And one example for this I will now show you. It's called convolutional coding. So it's a relatively, let's say, older type of forward error correction. You have here on the left the input, your data stream, and it's fed into a series of shift registers. So every time a bit enters here, everything goes one to the right. And then your output is actually not your input bit but a series of two symbols for each bit, each of them a result of a series of exclusive or operations here, depending not only on the input bit, but also on the state, so in your shift registers. And what this does is basically the information on a bit is smeared over a series of consecutive bits. So if you then get a problem in your reception, so you maybe lose data, you can still infer the content of your bit from the other bits you received correctly. But on the other side, for each bit your input here, you get two bits out. So you actually double the amount of bits you have to transmit and to counteract that, one technique would be a so-called puncturing where you have a fixed pattern of bits you just omit. So you can say, okay, now I have this factor two due to my convolutional coding, but I now just drop, let's say, every third bit or so and the receiver knows this, so it can just assume, okay, this is a 0.5, let's say, and then fix it in the decoding process. How can such a decoding process look? One example would be here, where to be decoding. This is using, I won't completely explain it here, but it's using the fact that within your shift register and an incoming bit, you only have certain allowed state transitions between the shift register. So for example, if you have a shift register state, which is all zero, you can just go into exactly two different states with the next bit, which is still all zero or one, one, and the others are zero. And from this, you get exactly two possible two bit combinations which are emitted from your convolutional coder. So in the other way around, if you receive a symbol, you can make a guess and the more symbols you get, the better guess it is. What kind of path the signal takes through your allowed state transitions? So in the end, after a certain amount, you are fairly certain that you know which kind of sequence you received. So what the effect is on your bit error rate graph is basically that you get the kind of coding gain through any kind of forward error correction. So when you look at the bit error rate graph without coding and with coding, you basically get additional energy within your signal which you don't have to really put into. So if you are further interested in different kinds of space communication, two things you might be able to duck that go. One is the CCSDS standard sets, which is basically a committee which exclusively deals with standards in space data exchange and communication. So there you can find information on the nitty-gritty details of how bits are encoded on satellite telemetry and telecommands. So let's say housekeeping data, temperatures, battery voltages and so from satellites or telecommands, so how to command the satellites. Quick word of warning, actually it's more like an option sheet for your communication. So typically for each mission, they select a certain amount of, I want to have this forward error correction at this bit rate and this modulation type, so it's not all, let's say directly compatible, you have to have some kind of additional knowledge to decode a satellite telemetry and telecommand stream. The other thing is the Etsy standards for DVBS2 is 2x, which is basically the broadcasting standard for satellite TV or a set IP if it goes over the geostationary satellite, so all these things you can find in these standards. And with this, I'm at the end of my presentation, so thank you for your attention and I'm open for questions. Yes, thank you very much for the interesting talk. We still have plenty of time, so if you have any questions that you would ask our speaker, please move to our microphone angel who is standing there as a living microphone stand. You just walk over there and you ask your question and we'd be very happy if anyone would feel ready to ask something because as I mentioned, we still have plenty of time. I'm hoping that this is a question that is moving here in the second row. Yes. Hi, thanks for the talk. I have a question regarding the sweeping for locking to the frequency. You said that the sweep is done on the transmitter side. Is this common? Why is this not done on the receiving side? Because the receiver would know when it loses track of the frequency. Yeah, actually it's usually done because communication with a satellite and the sweeping is typically done for operating a satellite, so telemetry, telecommand, not so much for, let's say, broadcasting satellites. So it's initiated by the ground station. So typically, the ground station has the much more complex systems so it can more easily react to changes in the, let's say, transmission stream. So it starts the transaction basically by starting the sweeping process and then what often happens is that the satellite maybe wasn't even transmitting before, but then when it sees this energy on its input band, it also starts transmitting itself. So just in case you didn't program it to transmit in time. So it's just done on the receiving side because that's more easily controlled by the ground operation crew. Thank you very much. Do we have any further questions? Don't be shy. It's not every day that you have a specialist for satellites in the room that you can ask pretty much anything. Nobody? Yes, please do move to the microphone. Thank you. Oh, and there's someone else. Okay, so, but you first, because I saw you first and then you, thank you. Sanctuary, do you know if there is a specific protocol for space to space communication or how protocol are space to ground or ground to space? Yeah, typically, for example, CCSDS standards has this whole option sheets of communication. And if you don't want to use the options for space to ground communication, for space to space communication, there are also some standards there you can look at. So for example, for advanced orbital systems or inter-satellite links. The next question from this microphone, please. What special encodings would you do if you have to transmit something really far away like to the new horizons probe or to Mars? So there you probably would use a code of code which is very close to theoretical optimum. So from information theory, you have a theoretical optimum you can get for a certain signal to noise ratio. And there you probably use turbo codes, for example. So I didn't present it here, but turbo codes are basically a set of encodings where you have on the encoding side two encoders working in tandem, so each producing a data stream. Then on the decoder side, you can use this to infer even more data by doing soft decisions on the encoder. Or another example would be, which is more, let's say, from the broadcasting side where they selected, I think also because the patent was lapsed at the point, was loaded into the parity check matrixes, but there you have the problem that you need very huge blocks. So for example, for satellite TV, typically the block size is about 65K bits, you have to transmit as a whole to get this high gain in the encoding side. So it's probably not optimal if you just want to send a couple of bits to your probe. Thank you very much. Is there a further question? No, is there a question from the internet? Also, no, this is your last chance. Any other questions? Yes. Hi. So what's the nature of the noise? Is it like across the spectrum? Is there any ways that you can make the signal more resilient to noise or interference? Yeah, so I didn't talk about interference because typically if you have, let's say, a professional ground station, you build it in a valley, so you don't get that much interference from, let's say, Wi-Fi access points or so. Actually, if you are designing a satellite system, often you assume, if you are building it for, let's say, broadcasting or point-to-point links, that it's about wide noise, so just equal across the spectrum. Of course, if you do something else, let's say a GPS-like system, so radio navigational satellites, you have to account for different kinds of noise or interference. For example, in urban environments, you get the multi-power propagation when you get a signal directly from the satellite and another one reflected from the Skype scraper next to you. And then you need different techniques to accommodate for that. Thank you. Further questions? I'm under the impression that we are running out of questions, so thank you very much for all your questions and please give another warm round of applause to our speaker. Thank you very much for answering all the questions.