Відео

I’ve also seen that statements in textbooks and I’m not particularly fond of it, mostly because it only makes sense for receiver processing and not transmission. The ZF combining vector can be scaled in different ways. What I assume in the video is that we normalize the vector so it has norm/length 1. This implies that the noise variance remains constant but we sacrifice signal power to eliminate the interference. Another approach is to scale the ZF vector so that the signal power remains constant. In that case, the price to pay for removing interference is that we amplify the noise. These two interpretations are equivalent in the sense that the SNR after ZF is the same. Which option that is closest to reality depends on the hardware implementation. LMMSE tries to maximize the SINR (which is equivalent to minimizing the MSE). There are two terms in the denomination: interference and noise. To mitigate interference, we need to sacrifice parts of the desired signal - the parts that lies in the same dimensions in the vector space. To “mitigate” noise, we just need to keep the desired signal power as strong as possible. LMMSE finds a balance between these things: A balance between ZF and MR.

Thanks! It is now much clear to me.

A lot of work on “massive MIMO” focused on making the algorithms universally applicable so they don’t require a particular propagation environment or array configuration to be used. All such results are directly applicable to both near and far field channels. New challenges arise when we want to exploit the specific channel characteristics in the algorithms to improve things like channel estimation. The broadcasting of information might also have to be done differently.

This is a good question that I tried to answer in a blog post a few years ago: ma-mimo.ellintech.se/2018/11/02/when-are-downlink-pilots-needed/ From a theoretical standpoint, it is sufficient to send pilots in one direction and then use it to make the effective channel predictable in the other direction. However, in practice, we often send pilots in both directions to deal with hardware impairments. Insufficient calibration between the hardware components used in uplink and downlink will lead to slightly different channels. This is particularly the case with phase drifts that are added with different signs in the uplink and downlink.

@@WirelessFuture Thanks for your explanation! In this blog, it is assumed that uses are equipped with single antenna and downlink pilots are not needed theoretically. For the case where multiple antennas are equipped, should we send downlink pilots? I think users must know h to combine signals ( MRC et al.) and hence downlink pilots are needed. Is that true? Additionally, if downlink pilots are needed, the number of downlink pilots is proportional to the number of antennas at the base station M, which may incur high signaling cost in massive MIMO, right?

You are right that downlink pilots will be needed whenever the user device has multiple antennas. In that case, one should transmit one pilot per downlink beam, and beamform/precode it as one will do what the downlink signal. The number of downlink pilot is then equal to the number of beams, which is typically much lower than the number of base station antennas. So the signaling code is not overly high - typically the same as the uplink pilot signaling.

@@WirelessFuture clear explanation, thanks for ur patient response!

The fact that one third of the worldwide mobile data traffic is transferred over 5G is an example of why 5G is very useful.

Yes, these traditional terminologies were defined for open-loop systems, which remains relevant for broadcasting of information. The main data transmission is often done in closed-loop mode nowadays and such systems are usually designed for fast fading and uses frequent pilot signal transmission to estimate/track the channels. We discuss this later in the course when we reach the “Massive MIMO” part.

@@WirelessFuturethanks for your explanation ❤

The general formula for the MMSE estimator (Slide 4) holds regardless of the distribution, but it is considerably more complicated to compute the estimate. I'm not aware of a general closed-form expression of it, but there are two special cases that have been treated in the prior literature: "Phase-aware MMSE estimator" and "Phase-unaware linear MMSE estimator". We cover these things in the following paper: arxiv.org/pdf/1903.07335

Spatial layers are data signals sent at the same time and frequency, but with different spatial directivity. It is like listening to two songs at same time - one with each ear. The capacity increases almost proportionally to the number of spatial layers. The word “almost” refers to that we need to divide the signal power between the layers and that there might be interference between the layers. With subarrays, we can reduce the latter losses since the spatial resolution of the transmission improves.

Yes, this is known as single-user MIMO or point-to-point MIMO. The number of beams that can be sent in this way is equal to the number of distinguishable paths between the transmitter and receiver. If you begin with sending one beam directly between them, then the next beam needs to be aimed toward a reflecting object that is located outside the first beam, from both the transmitter’s and receiver’s viewpoint.

@@WirelessFuture thanks for the answer, but what if there is no reflection path, it is simply LoS scenario

In a far-field free-space LOS channel, one can only transmit one beam since there is only one path. In practical LOS scenarios on earth, there are usually some reflected paths as well but they can be so much weaker than the LOS path that they doesn’t help much in boosting the data rate. This is why multi-user MIMO is particularly important in current and future systems because one can always send beams to multiple users, even in LOS scenarios.

@@WirelessFuture thanks for your very good explanation :)

When there is a need for more capacity, network operators have a choice between adding more base station sites or using a higher MIMO dimension at existing sites. The latter was not a real option until 4G LTE Advanced, and network densification has become less attractive since then. The fact that 5G mmWave networks have failed to become economically sustainable is an example of how further densification is unattractive. So yes, we have squeezed out the most we can from that. What remains is the battle between WiFi and 5G/6G in private networks, where the telecom industry hopes to make money from guaranteeing better performance than in WiFi.

@@WirelessFuture Thanks for taking time to answer this. Glad you clarified this aspect.

The telecom world mainly creates “bit pipes” and has managed to sustain an exponential capacity growth in their networks without increasing the prices at the same pace, so the mobile broadband business is quite successful. Other potential new use cases, focused on machine-to-machine communications, are growing slower than the telecom world has hoped for. But there is much investments into that right now so it might take off during the next five years. This is further discussed in the following video: 6G: What, why, and how? ua-cam.com/video/70BOKA0PmdE/v-deo.html

@@WirelessFuture I worked in the development of 5G back in 2017, I've heard it all before. Machine type communications, haptic communications, VR, robotic surgeries, etc. I was the first to demonstrate many of these on a fully virtualised prototype 5G system with function splits and ultra-low latency and jitter. Most of them are scenarios that either don't require the bandwidth/latency or they are likely to have access to fibre, or have no element of mobility so little reason to do over radio. I was vocal that it is BS but I was surrounded by celebrity professors and vendors who wanted to pump the BS hype, so I was hushed in favour of people who played the game and who took all the credit for my work in the end. As for the reason that prices have been the same is not because the telcos managed to keep the prices low but because connectivity is commoditised and nobody wants to pay a lot of money for it. Operators had tried to push prices a few decades ago and the churn went through the roof. They are attempting it again now by trying to monetise QoS with QoD. I'd love to see how that scales, considering that they are effectively proposing intserv as a solution to QoD. I guess all the decent network engineers have gone to hyperscalers and the telco is left with people who don't understand scalability of the control plane. They also don't understand traffic flows judging by the state of the CAMARA Project. The telco world has become a version of Boeing with business people at the helm instead of engineers. Anyway, rant over, thanks for your time and good luck.

@@WirelessFuture I was part of the 5G bandwagon back in 2017 and developed testbeds for it and showcased several world first use cases. I was vocal that most of the use cases are not based on reality, they do not consider whether enabling and supporting technologies are likely to reach the required maturity, and most importantly, they never investigate the cost/benefit over having a wired connection. Many of them do not even require mobility, and yet they were used to showcase low latency in 5G. But nobody listened to me because people were making careers out of selling snake oil in the hype. Over half a decade later and practically nobody cares about the ultra low latency or VR or the metaverse, or robotic telesurgeries. 6G is going the be a failure of even bigger magnitude than 5G. We are trying to invent convoluted ways to manage spectrum efficiently instead of investing on fibre and letting the radio handle cases of mobility. Fibre will always beat radio in performance and yet we are focused on mobile communications for use cases that don't need it, we spend millions over millions to solve spectrum management problems, and in the end we end up with a solution inferior to the performance of fibre. Is the whole telco world asleep at the wheel?

Yes, the transmit power at base stations is proportional to the bandwidth. The total power consumption also contains digital computations and losses in radio circuits. The latter things can also increase with the bandwidth and/or number of antennas, but are also improved by the gradual hardware improvements that happen all the time. I think the general goal of 6G will be to deliver much more data and better services with roughly the same total power consumption.

To my understanding, base stations are unlikely to have contiguous bandwidths that are larger than the operational bandwidth of an array. The problem rather arises due to carrier aggregation of multiple bands that are too far apart to use the same array. The industry has started to make products with so-called "interleaved arrays" where arrays for low and mid bands are integrated into the same box. As you pointed out, their design is intricate due to mutual coupling and other non-ideal electromagnetic effects, but it can apparently be done. Ericsson mentions one of its products on this page: www.ericsson.com/en/portfolio/networks/ericsson-radio-system/antenna-system/antenna/passive-antenna

@@WirelessFuture In the USA, the 6705 Ericsson radio is used for the 28 GHz band (considered mmWave), which can use up to 8 continuous 100 MHz carriers. Regarding interleaved arrays, a passive antenna with multiple ports is used, using the 4449 (new version 4490) low band radios and the 8843 (new version 4890 Ericsson) mid band radios. These configurations are widely used, using NR in 850 MHz and LTE in midband. For antennas, many brands are used, such as Commscope, JMA, etc. For C-band (NR), there are other antennas or radios from Ericsson considered, but they depend on which can be used, but there are like 3 options (2 antennas and 1 radio). I hope this helps.

WRC-27 is the last WRC before the product development and deployment of 6G will begin, so it is the last chance to assign new bands to IMT before that. In principle, any IMT band could be used for 4G, 5G, or 6G. But it is natural that a new generation uses a new band so it will not collide with legacy networks and will have access to substantially more bandwidth. When the ITU has finished the IMT-2030 requirements, there will be a submission period in 2027-2029 for RITs. We describe the timelines for 6G development in 3GPP and ITU in the following video: ua-cam.com/video/70BOKA0PmdE/v-deo.htmlsi=kqzkvrVkk9ZlbSBm

So, you like the video, yeah?

Yes, h2^H h1 is called the inner product or dot product between two vectors, and it is a measure of how similar they are. I don’t know what equations you are analyzing, but the basic matrix and vector equations for MIMO communications can be found in my open access book: www.nowpublishers.com/article/BookDetails/9781638283140

Commercial 5G networks are mostly supporting enhanced mobile broadband so far, which means delivering 4G-like performance but to many more users simultaneuosly and at a lower price per GB. Hence, most users might not experience a big difference compared to 4G. User devices such as mobile phones are not using more power as we improve the technology, but keep following the specific absorption rate (SAR) limits set by the regulatory bodies in each country. However, thanks to the beamforming enabled by having multiple antennas in the device, the signal can be focused toward the base station so that less reaches the user.

A cellular generation is first developed under 10 years and then deployed a refined under another 10-15 years. That is why 6G is being developed now even if 5G has only been partially deployed. You are right that many of the new anticipated 5G services remain to be commercialized, particularly those that go beyond “faster speeds” and “capacity for more users”.

It takes many years to roll out a new G, with significantly more advanced tech. Also, you're not forced to drop an existing G and then jump to a new one. My carrier still has 2G and 3G available, in addition to 5G. So not a lot of people are forced to run out and get a new phone. Even when that happens, carriers around here will offer a free replacement, though it won't be the latest and greatest. The older tech will be gradually phased out, until it's time to pull the plug on it.

The reactive near-field zone of an array of half-wavelength-spaced antennas is larger than the reactive near-field of an individual antenna. However, the considered subarrays are far outside each others’ reactive near-field zones so they will not affect each other or grow with the number of subarrays.

Thanks for watching and for noticing the typo!

@@WirelessFuture Why is your reply 5 hours earlier than the post you're replying to? Proactive? 🙂

If you write up the Lagrangian and compute the derivative with respect to \xi_k, you will get something of the kind: log2(1+(P*beta_k)/(xi_k*B*N0)) - log2(e)(P*beta_k/xi_k)/(B*N0+P*beta_k/xi_k) - Lagrange multiplier. This should be equal to zero for all k. Since the Lagrange multiplier doesn't depend on k, the expression can only be zero for all k if beta_k/xi_k=constant. It then remains to identify the right value of the constant.

The easiest thing is to use a normal mirror and rotate it so it reflects the signal in the right direction. I don’t know how to build something more RIS-like with such basic components. You can possible have multiple small mirrors and rotate them individually.

The ITU requirements refers to the performance (data rate, latency, etc.) that a technology must deliver to be called 6G. The 3GPP specifications refers to the details of the standard: How does the technology function in order to reach the performance requirements? ITU releases requirements once per decade (4G, 5G, 6G,…). 3GPP refines its standards every other year to add more functionalities, or making existing functionalities more attractive to implement in commercial products.

There is nothing in the pipeline. This was my latest talk on the topic, so I will likely create a new presentation during the autumn and then give it a few times in physical meetings before I record it. Is there something particular that you are looking for content on?

I’ve heard that the 3GPP channel models are being enhanced to also capture sub-THz bands. These bands could be used for fixed wireless backhaul links (or even fixed wireless access), where one can guarantee line-of-sight conditions in the deployment. However, I think sub-THz communications to mobile users is much further into the future, and will not happen until after mmWave communications to mobile users become a commercial success (if that will ever happen).

When I worked on my thesis, I spent several days trying to understand what scheduling algorithms were used in the latest standard. I then realized that the standard does not specify such things. Most things related to resource allocation is purposely not standardized so that the vendors can compete on making the most efficient implementations. The standard is only a skeleton that specifies the most essential things related to protocols, when to transmit signals and how, etc. I believe the same applies to pilot allocation. The possible pilot sequences are standardized and how you tell the user which one is assigned, but you don’t specify how the selected is made or what kind of estimation algorithm will be used when receiving the pilot. I don’t know what practical systems use, but I would guess it is some kind of greedy algorithm that has been fine-tuned using extensive simulations.

@@WirelessFuture Thank you very much for your reply. That explains a lot! I guess I will stop searching on those standardisation websites for a precise pilot assignment protocol!

There were competing cellular standards in the 1G, 2G, and 3G eras, but the telecom industry has since then realized the benefits of having only one global standard so that the same equipment can be sold everywhere. The standard is only a skeleton that enables interoperability between equipment developed by different vendors, which can still compete when it comes to the implementation details. Since standardization is done with unanimous decisions, there are many features in the standard that were added for bargaining reasons but that might never be utilized in practice.

The radar features in 6G will be far from military-grade radar technology because the power levels, waveforms, and deployment locations are much different. It can probably be used to detect vehicles moving along a street and their speed and direction, but it is typically limited to locations that are within line-of-sight to the base station. Typical use cases might be to help the internal communication protocols by predicting user movements, or to detect irregular movements in a factory havinga an indoor network. If an adversary wants to know what happens in a person's home, they need to hack into WiFi network or spy on data transmissions from user devices. Some countries are afraid of that equipment from some other countries might contain features that enables that, either today or in the future. By the way, the signal losses are very large in wireless communications, so massive MIMO would not be an effective "energy weapon". It is mostly the energy from your devices that effect your body, and since it is non-ionizing radiation it only transfer heat to the body. Nevertheless, there are regulations that keep the transmit power below specific thresholds that considered to be very safe.

This is a good point that we discuss in Section 2.4 of the book "Introduction to Multiple Antenna Communications and Reconfigurable Surfaces" (Download here: www.nowpublishers.com/Article/BookDetails/9781638283140) The short answer is that the channel capacity is achieved by encoding the data using codewords whose length goes to infinity. The proof builds on generating those codewords with Gaussian distributed entries. In practice, we don't do that, but instead, we create the codewords using a combination of a modulation format (e.g., BPSK or 16-QAM) and a channel code. This makes the implementation much easier, and with the right design, we can come close to the capacity.

@WirelessFuture Thank you very much for your reply. The book is great and contains everything I want to know. I have started reading it. Thank you for this contribution.

When an outage probability expression has the form constant*SNR^-M, we say that the diversity order is M. It is this number that determines the slope of the curve, as illustrated in Figure 5.13. There can be multiple transmission methods that achieve the same diversity order but anyway differ in performance because the SNR-independent constant factor is different. This happens when comparing (5.56) with repetition coding. (There is no Exercise 5.12c, but I hope my answer covers what you were looking for.)

What we refer to as non-orthogonal access is a MU-SISO scenario, so one serves multiple users but has only one antenna at the base station.

@@WirelessFuture Clear explanation,thank you! I noticed that there is a technology named 'NOMA (Non-orthogonal Multiple Access)'. Does multiple means base station has multiple antennas? If so, Does NOMA equals MU-MISO?

This is the correct result and interpretation. The point that we want to make with exercise is that the scaling factor matters when studying problems like this. (a) The scaling factor gives a zero variance asymptotically, so this is not the central limit theorem by the law of large numbers. (b) The scaling factor gives a finite variance, and this is when the random variable becomes Gaussian distributed according to the central limit theorem. (c) The scaling factor makes the variance go to infinity asymptotically, so there is no convergence to anything.