Annotation of the results obtained in 2023
We have proposed an improved spatial reuse method to service RTA traffic together with delay-insensitive traffic. It uses a scheduling algorithm that applies a greedy approach to solve the problem of minimization of channel access delay for stations with RTA traffic. At the same time, the scheduling algorithm does not deteriorate the fairness of resource allocation for stations with ordinary traffic and even improves the total network throughput. We have shown that the proposed method can almost halve the 0.999-quantile of RTA transmission delay compared to the standard spatial reuse method. We have developed a new method to service RTA traffic with several access points in several frequency channels with spatial reuse. The new method uses an algorithm to classify the stations with ordinary traffic but, unlike the single-channel method, considers the peculiarities of signal propagation in each channel. Moreover, we have developed a new scheduling algorithm that solves the combined problem of distributing stations among the channels and determining their transmission order. We have integrated the new method into the ns-3 simulation platform to analyze its efficiency. Simulation results show that the new method can almost halve the 0.999-quantile of RTA traffic delay compared with the method that does not take into account the multilink channel access. We have proposed a method to construct the precoder matrix while using MIMO and NOMA together. This method forms nonorthogonal spatial streams for users from different groups, keeping the orthogonality between the groups and completely eliminating inter-group interference. We used the simulation platform constructed during the Project to estimate the efficiency of the proposed method. Simulation results show that the proposed precoder significantly improves the mean geometric throughput and outperforms the approaches known from the literature for moderate signal-to-noise ratio values. It also can be used together with existing precoders for downlink NOMA-MIMO systems. We have developed the error model of reception of NOMA frames with arbitrary configuration, i.e., the distribution of power within the NOMA frame, signal-to-interference and noise ratio, and modulation and coding schemes can be arbitrary. The model has been integrated into the ns-3 simulation platform. We have used the model to estimate the efficiency of the common use of NOMA and MIMO and to check the accuracy of a popular approach to simplify the model of NOMA frame reception by changing the interference to the additive white Gaussian noise. We have shown that such an approach overestimates the signal-to-interference and noise ratio requirements to achieve the required reliability, which results in a lowered estimation of the use of NOMA and MIMO efficiency in Wi-Fi.

 

Publications

---

Annotation of the results obtained in 2021
We have studied Real-Time Applications (RTA) traffic service methods for Wi-Fi networks, aiming to develop a method that can satisfy the RTA traffic quality-of-service requirements (delay less than 1-10 ms with probability not less than 99.999%) and can maximize the throughput available for background traffic. We have developed and studied several RTA traffic service methods based on preliminary reservation of wireless channel and have compared these methods with each other and with the method based on tuning channel access parameters and limiting the transmission duration for background packets. The comparison of the methods has shown that the method based on tuning the channel access parameters has higher efficiency in case of intense RTA traffic (average frame interarrival time less than 50 ms). In other cases, the best efficiency is achieved with the method, for which a station reserves the channel and indicates the time interval when the actual packet arrival is expected, while the other RTA-stations can transmit their data if their transmissions do not intersect with the indicated time interval. This method can satisfy the requirements on the 99.999%-quantile of delay for RTA traffic and at the same time can achieve the efficiency of channel usage for background traffic approximately 20% higher than for other methods. We have developed and studied several RTA traffic service methods based on the usage of multilink channel access. We considered methods based on tuning of channel access parameters, on the usage of control channel and on mapping different kinds of traffic to different channels. The obtained results show that when RTA traffic consists of short frames, then the lowest 99.999%-quantile of delay for RTA traffic and the highest efficiency of channel usage for non-RTA devices are achieved when non-RTA and RTA traffic are mapped to different channels. When RTA frames are long and the RTA traffic has low intensity, the lowest quantile of delay is achieved with the method based on the control channel. When RTA traffic has high intensity and the RTA frames are long, the best efficiency is achieved with method based on tuning the channel access parameters in such a way that the RTA stations always win the contention for channel access with non-RTA stations, together with even distribution of stations over the available channels. We have considered one of the most important problems arising during the implementation of Uplink Non-Orthogonal Multiple Access (UL-NOMA) in Wi-Fi networks: the development of channel access method compatible with the random-access method used by legacy devices. We have proposed two channel access methods based on UL-NOMA: asynchronous and synchronous channel access methods. We used the simulation platform developed in the project to compare these methods with each other and with the legacy Wi-Fi channel access. The obtained results show that the usage of synchronous NOMA increases the total Wi-Fi network throughput and the geometric mean of the network stations’ throughputs by up to 100% in comparison with the asynchronous channel access and with legacy Wi-Fi channel access. We have developed a mathematical model of synchronous UL-NOMA based on discrete time Markov chains which can be used to estimate the aforementioned performance indices. We have also developed a mechanism for UL-NOMA reservation signal transmission and have found experimentally the parameters of reservation signals that enable orthogonalization of reservation signals at the access point. With using UL-NOMA in Wi-Fi, stations of the same network that experience different channel conditions can use the difference in the received power of their signals to send their frames to the access point at the same time on the same frequency. The access point can process such a frame superposition using the successful interference cancellation method. When two frames are transmitted at once, their preambles and pilot subcarriers overlap, as a result, the channel estimation and phase correction cannot be done independently for both frames. We have developed a UL-NOMA frame structure that solves this problem and is backward-compatible with the legacy Wi-Fi devices. We have developed a prototype of a Wi-Fi UL-NOMA transceiver and have found combinations of modulation and coding schemes (MCSs) for simultaneously transmitted frames for which the probability of reception is close to 100%. We have developed a simulation platform to model NOMA methods in Wi-Fi networks. The model includes a four-parametric error model which provides the probability of NOMA-frame reception depending on the NOMA-frame configuration. The model is based on experimental data. For DL-NOMA, we use the experimental platform developed by us before, while for UL-NOMA, we use the experimental platform developed in the project. In our model, we also use the dependencies of the numbers of received main and embedded frames on the signal to noise ratio for different MCSs used for both frames. We have shown that the results obtained with the developed platform match well with the experimental results known from the literature. We have also developed the models for NOMA channel access methods studied in the project and incorporated them into the simulation platform.

 

Publications

---

Annotation of the results obtained in 2022
We have developed and studied methods to serve Real-Time Applications (RTA) in Wi-Fi networks. We considered a problem to satisfy the quality of service requirements for RTA traffic (delay less than 1-10 ms with probability not less than 99.999%) and to maximize the throughput available for low priority traffic. Using the results of the extended research on RTA traffic service, we have developed an improved method of preliminary reservations, based on an algorithm of switching between the two previously proposed methods. We have shown that for efficient operation it is sufficient for the access point to know only the number of RTA stations in the network, duration of transmissions by ordinary stations, duration of RTA frames and reserving frames. We have shown that the new method can improve the efficiency of channel resource use for ordinary stations by more than 100% when the RTA traffic is predictable and has low intensity (less than 250 frames per second) and has strict requirements on the probability of RTA frame delivery (99.999%) with a specified delay limit (less than 5 ms). We have proposed and studied a method of RTA traffic service that uses several frequency channels. The method is based on RTA frame duplication and asynchronous contention for access in each channel. We have also proposed to improve this method by using priority channel access parameters. Comparison with previously developed multichannel methods has shown that the new method that duplicates frames shows the lowest 99.999%-quantile of delay in case of high intensity (average frame generation time by a station is less than 100 ms) of RTA frames with long duration (1 ms). We have developed a generalized mathematical model of data transmission in a wireless local area network (WLAN) that uses any of our developed multichannel methods. Our model can be used to find performance indicators both for RTA traffic and for the traffic which is insensitive to the delay. Validation results have shown that the model can calculate even 99.999%-quantiles of delay with high accuracy. These quantiles are needed to evaluate the efficiency of methods while serving RTA traffic with strict requirements on reliability. The developed model can be used to configure the algorithm at the AP to switch between the methods for RTA service and for throughput maximization for ordinary stations. We have proposed a channel access procedure for RTA service with several access points in WLANs. The procedure uses the mechanism of spatial reuse of channel resources, with which an access point can allow stations transmit RTA frames to other access points while it receives the data. To further evaluate and improve the efficiency of this procedure, we have developed in the NS-3 simulation platform a model of a heterogeneous network using the spatial reuse of channel resources for uplink traffic and a basic version of the proposed channel access procedure. We have developed an analytical model of WLAN that uses uplink (UL) non-orthogonal multiple access (NOMA) with reservation signals and RTS/CTS mechanism. Reservation signals and RTS/CTS mechanism provide backward compatibility with EDCA mechanism used in Wi-Fi networks. The developed model estimates the total throughput of devices in the network. The results show that the use of uplink non-orthogonal multiple access increases the total throughput by 30% compared to the EDCA mechanism. We have developed a prototype of a transceiver that uses UL-NOMA technology in Wi-Fi networks and have modified it to compare the noise resistance of Parallel Constellation Demapping and Successive Interference Cancellation methods when receiving NOMA-frames. Also, using the prototype, we have experimentally determined the channel conditions and Modulation and Coding Schemes for which the use of UL-NOMA increases network performance compared to existing Wi-Fi access methods in case of two transmitting devices. Also we have investigated the efficiency of NOMA methods in the uplink in Wi-Fi networks, and have compared it with other multiple access methods such as OFDMA and TDMA. We used simulation tools to take into account a large number of factors that affect the functioning of the wireless networks, such as various traffic types or inter-network interference. It is shown that if users are far from the access point, then NOMA loses to OFDMA because of the power spectral density increase due to channel narrowing when using OFDMA. Considering this result, we have proposed an algorithm for the joint use of NOMA and OFDMA. It is shown that the proposed algorithm increases the network throughput and reduces data transmission delays, using both the advantages of the OFDMA (using the heterogeneity of channel conditions by frequency, increasing the power spectral density) and the NOMA (multiplexing users with very different channel conditions). We have studied downlink NOMA in the presence of phase noise. To reduce the influence of phase noise, we proposed to rotate the signal constellation of a lower power frame of the NOMA frame. To find the rotation angle, we developed an analytical model for a NOMA frame reception that estimates the bit error rate for various phase noise values, AWGN and constellations of the NOMA frame. Obtained results were validated using the developed prototype of the NOMA transceiver with the ability to rotate the signal constellation of a lower power frame by a given angle. The experimental results show that the rotation of the signal constellation by 45 degrees increases the robustness. The gain in robustness is determined by the combination of signal constellations of the NOMA frame. The obtained results are especially important for stations located on the border of the transmission range of the central unit. Thus, when access point transmits data using the most reliable modulation and coding scheme, it is able to transmit data in parallel to stations with better channel conditions on higher order modulation and coding schemes. Also, we have evaluated the joint use of NOMA and MIMO in one and multiple streams. It is shown that the joint usage of NOMA and MIMO with beamforming on one device is effective for devices with similar channel characteristics. We have integrated the obtained NOMA frame reception error model into the common NS-3 protocol stack, which made it possible to apply a more accurate frame transmission model in the development of resource scheduling algorithms using NOMA. In particular, we proposed and implemented a radio resource allocation algorithm that uses the OFDMA and NOMA together. Based on the results of comparing algorithms, we concluded that the combined use of OFDMA and NOMA reduces delays in Wi-Fi networks compared to using only one of these methods. Also, we have modified the error model taking into account the previously obtained results. Moreover, we develop a module for loading channel characteristics generated in channel models with clustered delay lines.

 

Publications

---