Annotation of the results obtained in 2022
In 2022 the team obtained many results, which are highly important for the creation of extremely-high-throughput wireless local area networks. Among the main results, it is important to note the following ones. 1. MU-MIMO technology, as well as MU-MIMO transmission scheduling algorithms, plays a key role for providing high throughput. To study these algorithms, it is important to have a channel model that has low computational complexity and acceptable accuracy. The team investigated approaches to speed up the generation of the wireless channel parameters, that are based on modification of fast fading procedure. The proposed approach accelerates the channel generation process up to 5 times and provides results close to the original model. 2. We demonstrated that the maximum achievable throughput in a WLAN using the implicit sounding mechanism significantly depends on the amount of residual error after MIMO channel non-reciprocity calibration. Although the mechanism of implicit sounding can significantly reduce the overhead, and thus reduces the sounding period itself by several times, this mechanism is inherent in the distortion of information about the channel due to hardware imperfections. The requirements for the residual error are determined, under which an increase in throughput is achieved with respect to the use of the mechanism of explicit sounding with the optimal transmission period. 3. We developed and investigated a generalized MU-MIMO transmission scheduling algorithm that greedily maximizes a given utility function. We considered two types of functions as a utility function, that are appropriate to use for servicing web and video traffic and correspond to the PF and LAS service strategies. 4. We proposed and studied an OFDMA-based channel access method that uses direct links. It is shown that the use of this method maintains a high download speed via HTTP protocol, providing a satisfactory quality of service for the traffic of virtual reality applications in the same network. 5. We developed the first ever mathematical model that describes the operation of a heterogeneous Wi-Fi network with coexisting single-link and multi-link devices. The model is based on Markov chains. The accuracy of the model was validated by simulation. With its help, it is possible to evaluate the throughput of devices in such a heterogeneous network. 6. We investigated methods of multi-link channel access in vehicular IEEE 802.11bd networks. In particular, we studied the influence of various scenario parameters on the efficiency of the methods under consideration. We formulated recommendations for choosing a multi-link channel access method depending on the number of devices on the same and opposite road side, the traffic, which improves the traffic quality of service in IEEE 802.11bd intelligent transport networks. 7. We proposed and investigated new preamble puncturing algorithms in scenarios where two closely spaced Wi-Fi networks serve devices with different channel widths and with different types of traffic. Compared to the algorithms presented in the previous report, the new preamble puncturing algorithms eliminate long-term unfairness in the resource allocation between different networks in such heterogeneous scenarios. Thus, the algorithms provide the best performance in terms of geometric mean throughput for wideband devices and instantaneous throughput for narrowband devices in each network. 8. We investigated the support of quieting mechanism by modern Wi-Fi devices. Using the developed experimental testbed, we studied the accuracy of obeying quiet intervals for different interval durations, quiet interval offsets from the beginning of the beacon interval, interval periodicity, and support for several scheduled intervals for real Wi-Fi devices. The study shows that despite the declared support, all devices under consideration obey a single quiet interval with significant deviations from the standard, in particular, they do not stay silent for the full length of the interval. Moreover, devices do not obey more than one quiet interval, which can prevent the implementation of resource distribution methods with queiting as currently considered by the developers of the Wi-Fi 8 (IEEE 802.11be) standard. The obtained results are presented in 8 scientific papers accepted/published in 2022, including 3 articles in international journals of the first quartile (Q1) on Web of Science or Scopus, 1 in a Russian journal, the translation of which is indexed by Web of Science/Scopus, as well as 4 papers presented at conferences. In addition, they are presented in two invited reports at the Russian and international conferences. In addition, the project leader has defended his Doctor of Science Thesis.

 

Publications

---

Annotation of the results obtained in 2020
The increasing growth of the traffic in wireless local area networks (WLANs) requires the increase of their bandwidth. Among the existing approaches to solving this problem, the most promising are the use of wider frequency channels and the use of multi-antenna (MIMO, Multiple Input Multiple Output) and multi-user multi-antenna (MU-MIMO, Multi-User MIMO) transmissions. However, as studies show, an extensive approach to achieve a further many-fold increase in throughput, i.e., only by increasing the number of antennas and increasing the bandwidth, has several problems. These problems are especially notable in WLANs, e.g., in Wi-Fi networks, which use unlicensed frequency bands and random channel access methods. This problem is especially acute since WLAN technologies are most often used as the last mile when organizing Internet access. The random nature of the interference typical for such networks and a large overhead induced by the size of information about the channel state and the traffic presence reduces the efficiency of radio resource scheduling algorithms. Carrier sense-based channel access used in WLANs does not always allow using wide channels. Interference from other networks operating in the same radio frequency range increases the probability of unsuccessful transmission attempts and leads to the selection of modulation and coding schemes with a low nominal rate. As a result, in practice, the network throughput is significantly lower than the nominal rate stated in the standard. In this regard, in recent years, along with an increase in the nominal rate, research and development of new methods for increasing the efficiency of radio resource management algorithms in WLANs with a large number of multi-antenna devices and wide channels are of increasing interest. In the project, the team has developed a simulation tool of a WLAN that uses MU-MIMO and OFDMA technologies. The developed model has low computational complexity compared with the existing models and allows simulating data transmissions using OFDMA and MU-MIMO. In particular, to compute the physical layer of the model we used the MATLAB environment for preselected typical scenarios (the number of users, access points, their location). To obtain a reliable model of a wireless channel, taking into account frequency selective fading and multipath propagation, we used the Quadriga model which is a three-dimensional model of a channel with clustered delay lines (CDL). To simulate the time evolution of the channel characteristics when the device moves or the environment moves, we implemented the Generalized Method of Exact Doppler Spread (GMEDS), which allows obtaining weakly correlated channel characteristics in time. Next, we applied the obtained channel characteristics in a system network model. We used the NS-3 simulation environment, which enables developing algorithms for WLAN operation using OFDMA and MU-MIMO technologies. With the developed tool, the team has studied various algorithms for selecting a modulation and coding scheme (MCS) in a WLAN, as well as their interaction with the mechanism for changing the transmission power (OBSS PD), first proposed in the IEEE 802.11ax standard (Wi-Fi 6). Using simulation, we demonstrated that the usage of MCS selection algorithms available in the literature leads to a decrease in the WLAN performance when applied with the OBSS PD mechanism. To solve this problem, we developed a new MCS selection algorithm that uses a particle filter to estimate the signal-to-noise ratio at the receiver. This method allows the correct estimation of the change in signal strength due to the OBSS PD mechanism. With our simulation, we illustrated that the developed algorithm enables increasing the total WLAN throughput by 50% compared to the existing MCS selection algorithms. The team has also studied multi-link in WLANs. With the introduction of this technology into the IEEE 802.11be (Wi-Fi 7) and IEEE 802.11bd standards, in addition to using wide channels, devices will be able to exchange data over several narrow channels. Unlike channel bonding, the multi-link device, in order to gain medium access in a wide channel, does not perform a backoff procedure in one 20 MHz (802.11be) or 10 MHz (802.11bd) channel, but instead in each channel independently. Due to independent processes of medium access and flexible choice of transmission parameters in each link, multi-link technology allows increasing the network total throughput, reducing transmission delays, and enhancing reliability. Also, for the IEEE 802.11be standard, this technology opens up the possibility of simultaneous use of several links not only in the same band but also in different bands, for example, in the 2.4 GHz and 5 GHz bands, which is not possible with the current WLAN channel bonding methods. When the channel access processes work independently in each link, multi-link operates in the so-called asynchronous mode. However, its use is inefficient if there is interference between the links. Energy can leak from the transmitting link to the receiving link if the frequency channels are closely located. To avoid this issue, the synchronous mode is used, which is, however, less efficient. We studied the conditions where the asynchronous multi-link mode use is possible without compromising network performance. For the cases when the synchronous mode is preferred, the efficiency of various channel access methods for this mode was investigated. We also have evaluated the minimum spectral distance between two channels, at which the negative effect of inter-channel interference is negligible, and the multi-link device can use these channels independently. We carried out an experimental study using an SDR-based testbed using USRP-2944R. With this testbed, we implemented an asynchronous data exchange following the IEEE 802.11ac WLAN standard. During the study, we determined the boundary for the minimum spectral distance between the channels to achieve a 90% frame receive ratio depending on the interference power, modulation and coding scheme, channel bandwidth, and antenna location. In addition, we have discovered the effect of “indirect” interference in multi-link devices. Due to this effect, the transmission quality deteriorates at relatively large spectral distances and it was explained by the characteristics of the analog filter and digital-to-analog converter on the device. As for the IEEE 802.11be standard, we considered the case where there is a priori interference between the links, and it is necessary to use a synchronous multi-link mode. This mode requires coordinated channel access in all links, which can be implemented in various ways. In this task, we analyzed several methods proposed in the 802.11be Working Group using simulation. We proposed a classification for these methods, which divides the methods into ones with a single channel access process and multiple channel access processes. We compared these methods by the total throughput of the network, where both legacy devices and multi-link devices coexist. Next, for each channel access method, we studied the time spent by each device type in one of the following states: successful transmission, collided transmission, channel contention, waiting for channel access in the adjacent link, etc. For the IEEE 802.11bd standard, we carried out a comparative analysis of two-channel and single-channel media access methods. We analyzed the methods in terms of frame transmission delays and packet delivery ratio. They were compared in the highway scenario using the NS-3 network simulator. Moreover, we evaluated the impact of two-channel devices on the performance of legacy devices that use the single-channel algorithm. The team also studied the problems that arise in Wi-Fi networks due to interference. A testbed was developed to measure the carrier sense thresholds used by existing Wi-Fi devices. Using the testbed, the values of the ED (Energy Detect) threshold were obtained, which is used to detect high noise levels and interference in the channel. The measured values were used to validate the simulation models developed in the NS-3 simulator. By simulation in the NS-3 simulator, we have investigated power and carrier sense thresholds control methods to improve the performance of dense Wi-Fi. We have studied the dynamic mechanism called OBSS PD used in the Wi-Fi 6 standard. The mechanism allows setting occupancy power and carrier sense thresholds. We have studied OBSS PD performance in the scenarios with overlapping networks and find out problems connected with the interaction of the OBSS PD with algorithms of MCS selection algorithms used in the literature, e.g., Minstrel algorithm. We have shown that in some cases the use of OBSS PD can lead to a decrease in the Wi-Fi network performance. The team proposed and implemented in the NS-3 simulator the rules for selecting the duration of transmissions when the OBSS PD mechanism is used. Also, we have proposed a modification of the developed MCS selection algorithm to use information about the source of interference available when we use the OBSS PD algorithm. By simulation, we show that when the OBSS PD mechanism is used with the developed modifications, we can reduce the data transmission delay in Wi-Fi 6 networks by up to 2.5 times.

 

Publications

---

Annotation of the results obtained in 2021
Every year, the traffic in wireless local area networks (WLANs) grows, and this requires higher throughput in WLANs. Commonly, this issue has been solved using extensive approaches, such as the use of wider frequency channels and the use of multi-antenna (MIMO, Multiple Input Multiple Output) and multi-user multi-antenna (MU-MIMO, Multi-User MIMO) transmissions. The further development of these approaches is limited by physical constraints: limited bandwidth in a frequency band, the possible number of antennas on devices. These problems are compounded by the fact that WLANs such as Wi-Fi networks use unlicensed frequency bands and random channel access methods. These problems are especially acute since WLAN technologies are most often used as the last mile when organizing Internet access. Increasing WLANs’ performance under these conditions is especially important since WLAN technologies are most often used as the "last mile" when organizing Internet access. The efficiency of the WLAN operation is reduced by the random nature of the interference, the operation of a larger number of devices of different generations in the same network, and a large overhead induced by management signaling. Due to interference from other networks classical algorithms do not always allow using wide channels, and the probability of unsuccessful transmission attempts increases. The dynamic nature of the wireless environment impacts the signal quality when using MIMO and MU-MIMO technologies. As a result, the deployed networks have much lower throughput compared to nominal data rates from the standard. The efficiency of the WLAN operation may decrease due to the random nature of the interference. Due to interference from other networks, classical algorithms do not always allow the use of wide frequency channels, and the likelihood of unsuccessful transmission attempts also increases. The dynamic nature of the signal environment negatively impacts the quality when using MIMO and MU-MIMO technologies. As a result, in deployed networks, data transmission rates are lower than those stated in the standard. In this regard, along with the development of extensive increase methods, research and development of new methods for increasing the efficiency of radio resource management algorithms in WLANs with a large number of multi-antenna devices and wide channels are of increasing interest. The team has investigated the channel aging effect and its impact on the performance of WLANs that incorporate MU-MIMO technology. The channel aging occurs because the transmitter obtains channel state information only during the sounding procedure which produces a notable protocol overhead. We have found that channel aging reduces SNR up to 10-15 dB when using four spatial streams. We have proposed an MCS selection algorithm that takes channel aging into account when MU-MIMO transmissions are used. The proposed MCS selection algorithm combines linear prediction of SNR and adaptation to the statistics of received and missed frames which helps to effectively cope with the channel aging. As a result, the proposed algorithm notably increases Wi-Fi throughput compared to existing algorithms that do not consider channel aging. We have developed a simulation model of the Wi-Fi 6 network to evaluate the impact of promising approaches to increasing the performance of Wi-Fi 7 WLANs, such as channel estimation with explicit and implicit sounding, the use of OFDMA and MU-MIMO technologies. We have shown that if stations transmit only control frames using random access, and user data is transmitted using OFDMA after the request of an access point, the network throughput increases, and the average packet transmission delay decreases. Also, we have shown that the use of an implicit sounding allows increasing WLAN throughput up to two times compared with the currently used explicit sounding. The team has studied multi-channel transmission technologies in WLANs. Devices can exchange data over multiple channels using brand-new multi-link operation in the IEEE 802.11be (Wi-Fi 7) amendment and channel bonding in IEEE 802.11n/bd. To gain access to a wide channel, a device performs the backoff procedure in a single subchannel and then transmits a frame in the free adjacent subchannels. In multi-link operation (802.11be) a device accesses each channel independently using separate radio interfaces. When interfaces in a multi-link device operate in each channel independently, the device works in an asynchronous mode. However, this mode is ineffective in the presence of strong inter-channel interference. When frequency channels are too close, power leakage from the transmitting radio interface to the receiving one may occur. To solve this problem, the synchronous mode is used, in which the simultaneous transmission and reception are never used in different channels at any time. In addition, we have determined the feasibility of using asynchronous multi-link operation in modern mobile devices. We have developed a testbed based on modern mobile devices and access points. This testbed simulates two multi-link devices exchanging data using multi-link operation in asynchronous mode. We have shown that the use of asynchronous multi-link operation in smartphones is questionable, but it is possible in laptops and tablets, as their size allows the antennas to be spaced far enough. For Wi-Fi 7 networks, the team has considered the synchronous multi-link operation, used in presence of strong inter-channel interference. We have analyzed several synchronous multi-link channel access methods using simulation. For these methods, we have proposed a classification that distinguishes methods with one and with multiple channel access functions. These methods have been compared in terms of the throughput of multi-link and single-link devices, coexisting in the same wireless network, as well as in terms of the resource allocation fairness between the devices. As a result of the research, methods with one channel access function are found to be ineffective. Among other methods, two are promising. One of these methods, in comparison with the other, provides more throughput for multi-link devices, but at the expense of significantly reducing throughput for single-link devices. If one compensates for this imbalance in the channel resources distribution, then this method can become the most effective one among those considered. Next, a study of asynchronous multi-link operation has been carried out. In the NS-3 network simulator, a network model has been developed using this operation. The developed model has been used to identify and study the problem of the block acknowledgment window stalling that leads to a decrease in throughput. A solution to the problem has been proposed, allowing to achieve the network throughput proportional to the total width of the channels used. For IEEE 802.11bd networks, the team has conducted a comparative analysis of the three channel bonding techniques and one single-channel access method in terms of the portion of users unsatisfied with the frame transmission delay and the packet loss ratio. We have evaluated the impact of 802.11bd devices on the performance of legacy 802.11p devices. Furthermore, we have explored the impact of the contention window size on the performance of the channel bonding techniques. Based on these results, we have proposed a recommendation that allows selecting the most suitable channel bonding technique depending on the scenario. Besides, we have proposed a symmetric primary channel selection scheme that significantly increases the performance of the network. The team has also analyzed algorithms for the preamble puncturing method implemented in the 802.11ax standard and expanded in 802.11b. This method allows devices in a WLAN to transmit data in a wide frequency channel, puncturing subchannels with high levels of interference. In this task, the team has proposed several algorithms that select a puncturing pattern. We have compared the algorithms in terms of the throughput in each network and the geometric mean of the throughputs. We have shown that preamble puncturing algorithms achieve better or equal performance compared to algorithms that do not apply preamble puncturing. We have also proposed a new algorithm that selects a puncturing pattern to maximize the estimated throughput at the transmitter. For Wi-Fi 6 networks, the team has investigated the performance of dense networks using the OBSS PD mechanism. This 802.11ax mechanism allows devices to perform transmissions during interference from neighboring networks. We have developed a scheduling algorithm that takes into account the direction of transmission and the location of the source of interference, which makes it possible to increase the network throughput and overall performance. The proposed algorithm is implemented in the NS-3 simulator. Using this simulator, we have evaluated the network performance when using the proposed algorithm and compared it with the performance of widely used scheduling algorithms in IEEE 802.11. The team has studied the devices' capability of transmitting packets in strong-noise conditions. We have conducted a more thorough investigation of such capability, found by the team in 2020, that allows a device to transmit frames in the channel with noise having higher power than the Energy Detection threshold. The investigation is based on gathering and analyzing distributions of time intervals between consequent frame transmissions. Based on experimental data obtained, the team has proposed a model of sending frames and has calculated the parameters of the model for various devices. We have also developed an experimental testbed to collect the channel quality data. The testbed consists of two software-defined radios: a Wi-Fi transmitter and a four-channel receiver. The latter is equipped with an antenna array on a rotating platform, which can position the array at a predetermined angle relative to the transmitter. Using the developed setup, we have collected a dataset, consisting of channel information extracted from 300,000 frames in various scenarios for deploying a Wi-Fi network. The information consists of complex attenuation coefficients at each subcarrier, received signal power, and channel impulse response. All the data obtained are publicly available for the scientific community.

 

Publications

---