An international team, including researchers from the Russian Academy of Sciences Institute for Information Transmission Problems, HSE University, and Shanghai Jiao Tong University, analysed the approaches used to optimise the data transfer rates of TCP and QUIC protocols in high-frequency wireless networks. According to the scientists, cross-layer solutions provide the highest gains in data transfer rates. The paper has been published in IEEE Communications Surveys & Tutorials, one of the most influential international journals in the field of telecommunications, boasting an impressive impact factor of 35.6.
Wireless communication technologies are constantly evolving to offer users increasingly higher data transfer rates. Currently, Wi-Fi and mobile systems functioning within the conventional low-frequency bands (below 6 GHz) offer bandwidth capacities of up to several hundred megabits per second per user. Providing multi-Gbps data rates will be the next step. This bandwidth capacity will accommodate cloud gaming, virtual reality, and numerous other technologies.
It is possible to substantially increase data transfer rates in wireless networks by using high-frequency ranges, such as millimetre waves (from 30 GHz to 300 GHz), terahertz (from 300 GHz to 3 THz), infrared (from 300 THz to 420 THz) and visible light (from 420 THz to 750 THz). For example, recent advancements in Wi-Fi standards and fifth-generation cellular systems (5G) have already incorporated millimetre wave bands for data transmission. The IEEE 802.11 Working Group is currently in the process of finalising the IEEE 802.11bb standard, also known as Li-Fi, which broadens Wi-Fi range by using light waves (LW). High frequencies are considered to be potential candidates for diverse future wireless communication systems.
Currently, the two protocols used for transmitting large amounts of data over the internet are TCP (Transmission Control Protocol) and QUIC (Quick UDP Internet Connections). They determine the process of fragmenting data into packets, transmitting, and subsequently reassembling it.
These protocols establish the optimal data transmission rate between the client (user) and the server. A channel's bandwidth can be influenced by factors such as the number of active users, the quality of network equipment, the time of day, and other variables.
The TCP and QUIC protocols need to 'guesstimate' the optimal connection speed with every new transmission. This can be accomplished through trial and error or through measurements, as the protocol analyses the network's current state and uses this information to calculate the achievable data transmission rate. Both approaches require time to determine the optimal speed.
According to the researchers, here lies the potential vulnerability—the Achilles heel—of future generations of wireless communication. The greater the frequency, the more complexities arise in data transmission.
While conventional radio waves can effortlessly navigate around obstacles such as people and objects, an obstruction between the transmitter and receiver in millimetre wave frequencies can instantly diminish the data transfer rate from several gigabits per second to mere tens of megabits per second. The terahertz range and beyond can be easily impeded by a raised hand, leading to a significant deterioration in channel speed. Thus, high-frequency connections offer high speeds but are notably unreliable and pose major challenges when attempting to determine the optimal bandwidth.
A group of researchers from the RAS Institute for Information Transmission Problems, HSE University, and Shanghai Jiao Tong University explored several solutions that have been proposed to address the challenges stemming from the poor reliability of the high-frequency range. All examined approaches have been categorised into seven groups based on two primary criteria: the method employed for performance enhancement and the layer within the protocol stack in which the modification is implemented. The researchers have ultimately identified cross-layer solutions as the most promising avenue.
Routine online activities, such as making a video call, necessitate the activation of numerous protocols operating at various layers of the network. At the physical layer, bits are transformed into electromagnetic waves. At the subsequent layer, the channel access protocol comes into play to determine when a device can and cannot transmit data. For information to flow seamlessly beyond the access point, a network layer protocol and a routing mechanism need to select the optimal path for transmitting data from the source to the destination. Subsequently, information reaches a transport layer protocol, such as TCP, and from there, it travels up the layers to reach the user's applications.
Currently, each protocol operates within its layer and addresses only its specific and distinct tasks. For instance, TCP focuses on assessing the channel's bandwidth, while the algorithm at the channel layer responsible for the data transfer rate ignores the presence of TCP or any other protocol in a different layer. This results in a scenario where, instead of collaborating, each protocol operates in isolation from the others. This approach is extremely inefficient.
By facilitating direct interaction among various layers, cross-layer solutions effectively address numerous issues caused by the instability of high-frequency connections. Information from the physical layer, such as channel bandwidth, can be leveraged by transport layer protocols (TCP, QUIC) to enhance the management of data transmission.
The experiment demonstrates that implementing the XStream cross-layer communication protocol substantially increases file download speeds, bringing it close to the upper limit, which is the maximum attainable download speed under ideal circumstances (see the figure above).
With proper deployment and configuration, the solutions we have considered can significantly improve the performance of TCP and QUIC protocols in high-frequency wireless systems. These solutions differ in terms of the cost and complexity of their implementation. We believe that the most efficient of them are cross-layer solutions which can be employed either independently or in conjunction with other strategies to achieve the best outcomes.