Máximo Morales Céspedes, Borja Genovés Guzmán, Víctor P. Gil Jiménez, Ana García Armada
©SHUTTERSTOCK.COM/JAMESTEOHART
Vehicle-to-vehicle (V2V) communications are one of the most promising 6G scenarios. However, V2V communications demand high data rates and extremely low-latency requirements that are difficult to satisfy nowadays. Recently, vehicular visible light communications (V-VLC) have been proposed as a promising technology to guarantee such requirements. Multiple-input, multiple-output (MIMO) techniques can potentially achieve high data rates. However, when applied to V-VLC, they confront challenges such as the latency because of the need for closed-loop transmission or the correlation among optical channel responses. This article explains the use of blind interference alignment (BIA) for V-VLC based on the concept of a reconfigurable photodetector to provide linearly independent channel responses. It is shown that BIA solves the issues of MIMO techniques and may comply with the 6G V-VLC requirements. Besides, the vehicles are a convenient platform for implementing a reconfigurable photodetector. Finally, the open issues for BIA implementation in V-VLC are discussed to inspire future research.
Intelligent devices, autonomous cars, or smart unmanned aerial vehicles (UAVs) are some of the active elements in future vehicular communications, commonly known as the Internet of Vehicles (IoV). They demand ubiquitous mobile, ultrahigh-speed, and low-latency services, which are also considered critical challenges for future 6G wireless systems [1]. In this context, more efforts must be invested to lead a complete paradigm shift in future vehicular communications [2].
The performance requirements of vehicular networks cannot be met by 5G long-range communication technologies. In this context, dedicated short-range communications (DSRC) and VLC have been considered as proper technologies for enabling future vehicular communications. Besides, in intelligent transport systems, platooning applications with autonomous vehicles are expected to be a key 6G use case, where communication links are performed with vehicles located in the front and rear parts. In this way, line of sight (LoS) between the transmitter and receiver is highly probable, and DSRC and VLC perform well.
Since light-emitting diodes (LEDs) have been commonly adopted in vehicular systems, they can be used for converting the LEDs into a cost-effective technology to implement vehicular VLC (V-VLC). Wide and unlicensed bandwidths; controllable emission patterns; and affordable components are some of the advantages that make VLC a suitable technology for vehicular communications. For indoor applications, the coherence time of VLC is one order of magnitude larger than that for radio frequency (RF) systems [3], which reduces the overhead processing when updating the channel state information (CSI). However, the optical channel for vehicular communications is subject not only to transmitter and receiver movements but also to other parameters, such as atmospheric turbulence and weather conditions, which hamper the use of a channel estimation within a large time period [4].
The biggest challenges for V-VLC are interference management and the achievement of extreme low latency (below 0.1 ms for 6G systems). Thus, achieving high-capacity and reliable long-range communication links may be hampered by these unresolved issues. To solve the first one, MIMO VLC is proposed so that multiple optical sources are treated as different transmitters, generating multiple parallel data streams and maintaining the same total power budget [5]. The second issue may be solved by novel cooperative schemes requiring more simple decoding techniques. Although MIMO techniques have been developed for RF communications during the last decade for achieving both spatial diversity and multiplexing gain, their application to V-VLC is not straightforward, and several issues must be faced as described next.
MIMO precoding techniques typically require accurate CSI at the transmitters (CSITs). This requires feeding the estimated channel back to the transmitters, which generates a feedback delay, as shown in Figure 1(a). For VLC, the feedback channel is typically carried out through RF or infrared links, which increases the complexity and cost of the devices. Moreover, the mobility between vehicles, i.e., the longitudinal and lateral shift, generates channel variations more frequently than in indoor scenarios [6]. Therefore, the feedback delay in closed-loop transmission may lead to outdated or stale CSI, which causes the poor performance of MIMO precoding techniques. Open-loop transmission and short block-length theory have been proposed for achieving low latency in vehicular RF systems in [7] and [8], respectively. To the best of our knowledge, these approaches have not been considered for V-VLC yet. In this sense, a latency of around 1 ms is shown to be achieved for V-VLC in [9], which is above the 0.1 ms proposed for 6G.
Figure 1 Issues for implementing transmit precoding schemes that can be potentially solved by BIA in vehicular environments. (a) A closed loop for obtaining CSIT in MIMO precoding techniques (top) versus open-loop transmission for BIA (bottom). (b) The need for coordination among transmitters, e.g., UAVs providing connectivity to vehicles.
The number of uncorrelated paths in VLC is not as high as in RF because of its real nature and the lack of small-scale effects (the photodiode size is much larger than the optical wavelength). As a consequence, MIMO techniques are penalized because of the high correlation among channel responses. This issue is even more challenging in V2V communications based on VLC. Considering the front lights of a vehicle, the optical channel is more correlated as the distance between vehicles increases, i.e., both lights can be considered to come from the same transmitter.
Differently from RF, VLC depends on the availability of a LoS link as light does not go through opaque objects. However, V-VLC scenarios have moving elements; they include narrow and directional light beams, and the field of view (FoV) of the photodiodes is commonly narrow to minimize interference. These V-VLC features may jeopardize the LoS transmitter–receiver link, and V-VLC designs must consider these issues to make the communication reliable.
In this article, the use of BIA techniques for enabling 6G V-VLC is proposed based on the concept of a reconfigurable photodetector, which is discussed later. BIA techniques do not require CSI knowledge at the transmitter side for achieving multiplexing gain while using a very simple decoding technique. Thus, they fulfill the requirements of road safety applications, demanding low latency and reliable communication links. At the same time, BIA exploits the use of multiple transmitters and receivers, which is in accordance with the ISO 26262 “Road Vehicles–Functional Safety” standard establishing that every vehicle safety system cannot rely only on a single sensor.
V-VLC has always been considered as a fundamental use case in multiple VLC-related standards: IEEE 802.15.7, IEEE 802.15.13, and IEEE 802.11bb. However, most of the current standardization efforts have been focused on the exploitation of well-known RF chipsets to be adapted to light-fidelity (Li-Fi) systems. Besides, a Li-Fi ecosystem called Light Communication Alliance has been recently created involving agents in the full chain to guarantee a widespread deployment of Li-Fi solutions. However, a dedicated study of V-VLC to address the issue of that particular scenario is required, and it would fill the gap between academic research and industry. The light communication IEEE 802.11bb Standard is expected to be concluded in 2022. Considering that a new mobile communication standard is conceived every decade, we foresee that the first group activities on 6G standardization will start in around 2026. By that time, deployments and use cases based on the IEEE 802.11bb Standard will have matured, and a new light communication group can harmonize new Li-Fi-based 6G scenarios such as V-VLC.
BIA was initially proposed as a signal processing technique for achieving growth in degrees of freedom (DOFs), which can be interpreted as the multiplexing gain or simply the number of parallel channels transmitted free of interference as the number of users increases without the need for CSIT [10]. The implementation of BIA requires receivers able to switch among a set of linearly independent channel responses, denoted by preset modes. Then, a data frame following a specific switching pattern is generated to allow the interference subtraction, as shown in Figure 2.
Figure 2 A BIA example. The two front lights transmit data to a vehicle equipped with two reconfigurable photodetectors. The symbol transmitted by LED m to user k is denoted by ${u_m^{[k]}}$. Symbols ${u_1^{[1]}}, {u_2^{[1]}}$ and ${u_1^{[2]}}, {u_2^{[2]}}$ are intended to denote reconfigurable photodetectors #1 and #2, respectively, which are allocated in the same car. The mode selected by each reconfigurable photodetector corresponds to a color, either green or yellow. Mode switching during three time slots generates the resulting BIA pattern. This pattern can be repeated or expanded to adapt the frame transmission to the latency or coherence time requirements. TX: transmitter.
For illustrative purposes, let us consider a scenario comprising two transmitters, e.g., the two front lights of a vehicle that each sends a data stream to another vehicle equipped with two reconfigurable receivers, i.e., a reconfigurable photodetector for VLC. The period for data transmission comprises three symbol extensions, which correspond to three time slots in the time domain. Thus, the reconfigurable receiver of each user follows a pattern as shown in Figure 2.
Focusing on reconfigurable photodetector #1 in Figure 2, to which the symbols ${u_1^{[1]}}$ and ${u_2^{[1]}}$ are addressed, the interference because of transmission to reconfigurable photodetector #2, i.e., ${u_1^{[2]}} + {u_2^{[2]}}$ can be measured in time slot 3. Since this interference is received by reconfigurable photodetector #1 in the same mode as in time slot 1 (mode 1, associated with the green color in Figure 2), in which simultaneous transmission occurs, the interference can be subtracted from the signal received in time slot 1. Thus, the transmission of ${u_1^{[1]}} + {u_2^{[1]}}$ is contained in two modes of reconfigurable photodetector #1, which correspond to linearly independent responses. Therefore, the two symbols, ${u_1^{[1]}}$ and ${u_2^{[1]}}$ are decodable by solving the corresponding equation system. Similarly, the symbols ${u_1^{[2]}}$ and ${u_2^{[2]}}$ can be decoded by reconfigurable photodetector #2 following the same procedure. Therefore, four symbols are decodable during three time slots, i.e., 4/3 DOFs are achievable using BIA for the considered scenario. At this point, notice that other transmission schemes without CSIT, e.g., orthogonal resource allocation, are constrained to achieve 1 DOF.
For a broadcast channel composed of M transmitters and K users, BIA achieves ${(}{MK}{)} / {(}{M} + {K} - {1}{)}$ DOFs, comprising a specific switching pattern of preset modes, i.e., time slots in the time domain [10]. Note that achieving this performance requires a reconfigurable receiver providing at least M linearly independent channel responses. It is demonstrated that BIA obtains the maximum achievable DOFs without CSIT. Furthermore, other alternative BIA schemes are also optimal for partially connected or cognitive networks.
Since VLC are inherently frequency division duplex with the uplink and downlink operating at different wavelengths, obtaining CSIT requires a closed-loop procedure, as described in Figure 1(a), which penalizes the latency of the system because of feedback delay. Moreover, the lack of quality in the obtained CSI may lead to the poor performance of the MIMO precoding techniques. BIA allows us to get rid of this closed loop.
Getting rid of the closed loop allows us to transmit the signaling and data within the same frame, as shown in Figure 2. That is, the intended data in each frame can be decoded without the need for previous or following frames, avoiding feedback delay. Therefore, coherence time and latency requirements mainly depend on the frame length. In this sense, the resulting BIA pattern can be repeated or expanded to adapt the frame to these requirements. For instance, assuming a sampling rate equal to 10 megasamples per second and eight samples per symbol, the BIA pattern described in Figure 2, which comprises two symbols for estimating the CSI at the receiver (recall that it is not fed back) and three for data, is transmitted in 4 μs Moving to a more realistic application, expanding or repeating the pattern 30 times (see Figure 2) and considering that 10% of the frame is assigned to signaling, i.e., nine time slots for signaling and 90 for data, the transmission of the frame comprises 79.2 . On the other hand, the latency for closed-loop schemes also depends on the feedback delay, which hampers achieving low latency below 0.1 ms.
Achieving high data rates requires not only exploiting the benefits of MIMO, assuming a single access point made of multiple transmitters, but also spatial MIMO based on coordination among transmitters, as shown in Figure 1(b). BIA does not require data sharing among transmitters, and the cooperation is limited to synchronizing the vehicles in time, i.e., pairing two or more vehicles to create a bigger access point covering a wider area of the road.
Vehicular communications are typically LoS. As a consequence, the channel response corresponds to a frequency-flat fading channel. However, the rate achieved by MIMO precoding techniques is very sensitive to correlated channel responses, and then, the predominant LoS channel may handicap the implementation of precoding techniques in vehicular communications. On the other hand, the performance of BIA does not depend on the correlation among the channel responses of the users but on the correlation among the preset modes of the reconfigurable receiver of each user, which is in fact a design parameter that can be easily handled.
Motivated by the benefits of BIA, its implementation for VLC is first proposed in [11]. Interestingly, it is shown that BIA inherently satisfies the transmitted signal constraints for VLC, and in contrast to closed-loop precoding schemes, a dc bias depending on the transmitted signal is not required. Thus, the entire optical power range of the emitters can be exploited while maintaining their initial illumination function. Furthermore, it is demonstrated that by deploying photodiodes in angle and lens diversity, connected to a single signal processing chain through a selector, the receiver can modify the optical channel among a set of linearly independent responses. This receiver structure, referred to as the reconfigurable photodetector from now on, is described in Figure 3. Notice that the orientation angle has a direct impact on the incidence angle of each photodiode, which contributes nonlinearly to the resulting channel as a cosine function. The lens diversity also generates nonlinear responses as distinct geometries, e.g., hemispherical, coated, or truncated at different angles, are considered.
Figure 3 The architecture of the reconfigurable photodetector. It is composed of multiple photodiodes oriented in such a way that guarantees a distinct incidence angle for each photodiode, connected to a single signal processing chain through a selector. Each photodiode can be also equipped with a filter plus a concentrator lens.
However, the Lambertian channel model for indoor scenarios is not valid for dynamic outdoor V-VLC since it does not consider crucial effects such as the weather conditions; reflections on the road; or the lateral shift between vehicles. In [12], a channel model for V-VLC is provided combining the attenuation and geometrical losses, which result from the scattering/absorption and the transmitter beam spread, respectively. The different weather conditions, such as clear weather, rain, or fog, are modeled by applying correlation factors in the closed-form expression of the resulting path loss. Based on this model, a more practical channel is derived in [6] considering the specific parameters of vehicular scenarios, such as the width of the road and each lane; the probability of blocking; or the random lateral shift between vehicles.
It is worth noting that those models for V-VLC still depend nonlinearly on the incidence angle and the lens response. Therefore, the reconfigurable photodetector based on angle and lens diversity can be applied to V-VLC. Furthermore, beyond the need for reconfigurable photodetectors for implementing BIA, reconfigurable photodetectors are also useful to uncorrelate the channel responses among users for precoding techniques as proposed in [13].
Deploying multiple photodiodes pointing to distinct orientations in an angle diversity arrangement generates different incidence angles, which contributes to the optical channel in a nonlinear manner. In contrast to the angle diversity receiver concept proposed in [14], the reconfigurable photodetector exploits the angle diversity of multiple photodiodes deployed around the vehicle structure without the need for deploying them adjacent to each other. Moreover, the use of a selector to switch among the photodiodes instead of considering a signal processing chain per photodiode leads to reducing the receiver complexity.
Another way of obtaining linearly independent channel responses is exploiting the diversity between filters and concentrators, whose gain depends on the incidence angle in a nonlinear manner: for instance, using coated and spherical lenses, as shown in Figure 3. In this sense, the gain of the optical filter plus concentrator is \[{g}{\left({\varphi}\right)} = \frac{\mathop{\int}\limits_{S_0} {T}{\left({\theta_0}\right)}{T}{\left({\theta_1} \right)}{T}{\left({\theta_2}\right)}{T}{\left({\theta_3}\right)}{dS}} {\mathop{\int}\limits_{S_0} {\cos}{\left({\theta_0}\right)}{dS}} \tag{1} \] where ${S_0}$ is the integration area over the surface for which light passing through eventually hits the photodiode, and ${T}{(}{\theta}{)}$ is the gain function of each step inside the lens described in Figure 3. It can be seen that lens design provides multiple possibilities for obtaining linearly independent channel responses.
Nowadays, vehicles contain LED lamps and photodiodes that can be used for communications, too. We do not need to redesign our cars, motorbikes, trucks, or any other vehicle. The structure of the lamps in vehicles typically covers a wide angular range, e.g., headlamps usually cover not only the front, but they are curved to the lateral sides and inherently provide a set of filters and lenses, as shown in Figure 4. Enough angle and lens diversity can be achieved by exploiting this structure, guaranteeing linearly independent channel responses. It is worth noting that the integration of photodiodes close to the lights must include a barrier structure to avoid self-interference. Besides, additional photodiodes can be deployed with a minimum impact on the electrical, electronic, aerodynamic, and aesthetic aspects.
Figure 4 An arrangement of photodiodes in angle and lens diversity exploiting the vehicle’s structure.
Figure 5 A V2V communications scenario. BIA allows the transmission of multiple streams to the same vehicle or multiple vehicles. Moreover, assuming that cooperation is limited to time synchronization, i.e., no data sharing, BIA can be exploited to generate a multiple V2V system.
When a communication link is established with the front part of the vehicle, linearly independent channel responses can be guaranteed, taking into account the different elements. Allocating multiple photodiodes inside the headlamps can be distributed within a wide range of angles in both horizontal and vertical planes, which likely guarantees linear independence among the channel responses [15]. Moreover, there are two blocks of lamps separated by more than 1 m, adding more diversity to the communication channel. We also have the possibility of exploiting other structures, such as the fog lamps, the side mirror structures, or the car brand symbol, for allocating additional photodiodes. On top of that, most of the medium-high-range vehicles allocate photodiodes and cameras on the surface of the windshield glass for security reasons. Although these photodiodes typically provide a low bandwidth, specific photodiodes for VLC can be integrated into this structure.
There is also enough diversity in the rear part of the vehicle thanks to its architecture and number of elements. There are tail lamps that, as in the case of headlamps, are curved and cover both sides of the vehicle partially. We also find the back license structures; a separate brake lamp on top of the back window; and turn signals.
Everything mentioned previously is even more valid for the case of trucks because the number of lamps (i.e., the structures in which to deploy photodiodes in angle and lens diversity); the sizes; and the positions are wider than in usual cars. In the case of motorbikes, the location is a bit more complex due to size constraints. They could be replaced by a set of photodiodes placed on the helmet to exploit angle diversity. Notice that current helmets have communication systems for comfort, such as music or hands free. However, we foresee that, in the near future, novel augmented reality, such as virtual reality scout or driver-assistance systems, will be incorporated for safety reasons, which may require V-VLC.
Once the angle diversity has been shown to be valid in the communication between vehicles, we can also think about communications between vehicles and UAVs. This scenario is especially useful in cities and large roads. As can be seen in Figures 1(b) and 4, the link between the UAVs and the vehicles with diversity angle receivers can be easily established using the extra brake lights; the placement of the front camera; and the upper part of the headlamps and taillights. On the UAV side, the normal structure of UAVs (plane shape or spider shape) offers the perfect location for allocating photodiodes with enough angle and lens diversity to implement the concept of a reconfigurable photodetector.
The proposed vehicular scenario is depicted in Figure 5. We consider V2V communications so that the two front LED lights of the vehicle transmit two data streams to a vehicle equipped with reconfigurable photodetectors, similar to that described in Figure 2. A simple photodetector configuration is considered in which nine photodiodes following a geometrical arrangement, comprising an azimuthal angle equal to [±40°, ±30°, ±20°, ±10°, 0°] pointing parallel to the road, are deployed around the car. For the sake of simplicity, the use of lenses is not considered.
The considered vehicle LEDs satisfy Regulation 48 of the Economic Commission for Europe of the United Nations and the U.S. National Highway Traffic Safety Administration with a luminous flux equal to 6,000 lm, which corresponds to an optical power of about 24 W assuming a typical luminous efficiency of radiation. All other parameters used for our simulations are:
Note that clear weather is considered for the sake of simplicity. Adverse weather effects such as rain or fog would penalize the sum rate of all the transmission schemes similarly.
The achievable sum rate obtained by BIA, zero forcing (ZF), maximum ratio combining (MRC), and orthogonal resource allocation in a V2V communication is evaluated in Figure 6 as a function of the distance between vehicles. First, assuming perfect CSIT and that both vehicles are in the same lane, it is worth noting that the performance of ZF decreases abruptly as the vehicle moves away. This effect is given by the high correlation between the channel response of the photodiodes located in the receiving vehicle. Indeed, this correlation is more noticeable with the distance between vehicles since the two front LED lights of the vehicle can be considered as a single optical channel when the distance increases. On the other hand, the implementation of a reconfigurable photodetector in vehicles allows us to improve the achievable rate of ZF by selecting the photodiode that maximizes the performance of ZF (ZF reconfigurable), as proposed in [13]. Moreover, orthogonal resource allocation achieves a lower sum rate in the whole distance range. MRC outperforms BIA only for long distances, i.e., at low sum-rate values, since it exploits the channel diversity provided by the reconfigurable photodetector for maximizing the signal-to-interference-plus-noise ratio of the link instead of achieving multiplexing gain.
Figure 6 An achievable sum rate for ZF precoding, BIA, and orthogonal resource allocation as a function of the distance between vehicles.
Maximizing the DOFs requires closed-loop transmission, e.g., in this case, ZF precoding achieves ${\min}{(}{M},{K}{)} = {2}$ DOFs. On the other hand, BIA achieves the optimal DOFs without CSIT, ${(}{MK}{)} / {(}{M} + {K} - {1}{)} = {4/3}$, although still below the DOFs achievable by closed-loop schemes. Since the DOF metric is primarily concerned with the limit when the total transmission power goes to infinity, i.e., as the distance between vehicles becomes shorter, it can be seen that ZF outperforms BIA at very short distances with a greater sum-rate slope, which is given by the influence of the DOFs. However, the optical channel is more correlated as the distance between vehicles becomes larger. This correlation makes the sum rate decrease abruptly for ZF. On the other hand, since BIA does not depend on the correlation among the channel responses of the users, it obtains a greater sum rate in a wide range beyond short distances between vehicles.
If the V2V communication occurs between vehicles that are located in different lanes, it can be seen that ZF is considerably penalized, and indeed, the use of photodiode selection [13] is mandatory to achieve useful sum rates. On the other hand, BIA achieves an acceptable sum rate in a wide range of distances between vehicles. It is worth noting that the performance of both ZF and BIA can be potentially improved by considering more preset modes to the reconfigurable photodetector of the vehicles, i.e., adding more photodiodes pointing to other directions or using lens diversity. Furthermore, assuming synchronization based on pairing two vehicles, the performance of the multiple V2V system outperforms the sum rate achieved by V2V BIA transmission.
The latency depends directly on 1) the reference signals required for achieving CSI at both the transmitter and receiver sides (i.e., the larger the number of reference signals, the larger the frame, and as a consequence, the latency increases) and 2) the size of the data packet. As represented in Figure 7, the number of reference signals in MIMO precoding increases with the number of users since each user must feed the estimated channel back to the transmitter, and afterward, the specific precoded pilots are transmitted for each data stream. On the other hand, open-loop transmission based on BIA requires only a single reference signal per transmitter for achieving CSI at the receivers. However, as represented in the right-hand y-axis of Figure 7, the length of the switching pattern grows exponentially with the number of users and transmitters. As a consequence, current BIA schemes achieve extremely low latency for configurations with a reduced number of transmitters and receivers, which might satisfy the challenge pointed out for 6G wireless systems.
Figure 7 The number of reference signals for MIMO precoding and BIA and the length of the BIA switching pattern as a function of the number of users.
As introduced in the “BIA: An Overview” section, BIA does not require CSIT at the expense of using more time slots to cancel the interference. The number of time slots needed depends on the number of different transmitters. Besides, these slots must be within the coherence time to guarantee that the interference is aligned and can be cancelled. In scenarios where there is high mobility, i.e., the coherence time is short, BIA schemes cannot be that promising because the channel changes too fast so that the light cannot be aligned on time. In these scenarios, it would be interesting to explore what we call here wavelength BIA. By using the same example of the “BIA: An Overview” section with two users and two transmitters, in which three time slots are required, we can substitute the three slots with three different colors in red-green-blue (RGB) LEDs so that transmission is sent at the three different wavelengths. In this way, only one time slot is needed. In this sense, alternative reconfigurable photodetector architectures considering color filtering must be designed.
For a more complicated scenario with more users and/or transmitters, a combination between time slots and colors may be needed, but there will still be a reduction in the coherence time constraints. We could leverage commercial tricolor (RGB) or quadricolor (red-green-blue-yellow) LEDs, and the target color can be obtained by the corresponding weighted combination of those colors.
The fundamentals of BIA are based on a reconfigurable photodetector that can align the interference without the need for CSI knowledge. Photo-sensitive devices, already installed in vehicles, such as LEDs, could be exploited as photodetectors, i.e., employing LEDs as both transmitters and receivers would create a full-duplex bidirectional channel, and the wavelength selectivity of the LEDs would get rid of extra optical filters or barrier structures. Specific photodiodes for VLC are not typically installed in vehicles, but their low cost, small form factor, and affordability make their installation viable, without meaning an extra effort for the automotive industry. To enable BIA in V-VLC, some photodiodes must be installed in vehicles, and their arrangement must be done considering the characteristics of the vehicular environments.
Several BIA schemes have been developed for cellular or cognitive scenarios. Thus, motivated by the benefits of BIA applied to V-VLC, specific BIA schemes must be developed to optimize the achievable rate and latency subject to reliability conditions, such as the low probability of error considering specific modulation/coding designs and management of the frame structure. Moreover, the introduction of machine learning techniques in V-VLC [4] opens the door to dynamically adapting the BIA scheme as the conditions of the vehicular environment change.
The challenges of 6G wireless systems are even more difficult to address for vehicular communications, in which VLC is proposed as a promising technology. This article has discussed the implementation of BIA in vehicular communications based on the concept of a reconfigurable photodetector. In this sense, BIA allows us to get rid of the closed loop and to remove the negative effects of the channel correlation of MIMO precoding techniques while providing multiplexing gain. This leads to ensuring a high data rate and an extremely low latency required by 6G. We have also discussed the availability of light diversity in current vehicles and its benefit for BIA in VLC. Finally, we have introduced the open research challenges on the implementation of BIA in V-VLC, which must be solved for developing these novel transmission schemes oriented to vehicular environments.
This work was partly funded by Project “IRENE” (PID2020-115323RB-C33) (MINECO/AEI/FEDER, UE) and Project GEOVEOLUZ-CM-UC3M. In addition, this work has been partially funded by the Juan de la Cierva Formación (Grant FJC2019-039541-I/AEI/10.13039/501100011033), granted to the author B. Genovés Guzmán, and the Juan de la Cierva Incorporación (Grant IJC2019-040317-I), granted to the author M. Morales Céspedes.
Máximo Morales Céspedes (maximo@tsc.uc3m.es) is a postdoctoral researcher with the Department of Signal Theory and Communications, Universidad Carlos III de Madrid, Leganés 28911, Spain. His research interests include visible light communications, interference management, and signal processing applied to wireless communications. He has published more than 30 refereed articles, and he holds one patent. He serves on the Editorial Board of IEEE Communications Letters and the Open Journal of the IEEE Communications Society. He is a Member of IEEE.
Borja Genovés Guzmán (borja.genoves@imdea.org) is a postdoctoral researcher at the IMDEA Networks Institute, Leganés 28918 Madrid, Spain. He received his Ph.D. degree from the University Carlos III of Madrid in 2019 (Extraordinary Ph.D. Award). His research interests include Li-Fi and next-generation wireless networks. He was a recipient of FPU and Juan de la Cierva grants from the Spanish government. He is a Member of IEEE.
Víctor P. Gil Jiménez (vgil@tsc.uc3m.es) is an associate professor with the Department of Signal Theory and Communications, Universidad Carlos III de Madrid, Leganés 28911, Spain. His research interests include signal processing for wireless communications. He has published more than 30 refereed articles, and he holds one patent. He served as the IEEE Spanish Signal Processing and Communications Joint Chapter chair from 2015 to 2022. He is a Senior Member of IEEE.
Ana García Armada (agarcia@tsc.uc3m.es) is a professor with the Department of Signal Theory and Communications, Universidad Carlos III de Madrid, Leganés 28911, Spain. Her research interests include signal processing applied to wireless communications. She has published approximately 150 refereed articles, and she holds four patents. She serves on the Editorial Board of IEEE Transactions on Communications and the Open Journal of the IEEE Communications Society. She is a Senior Member of IEEE.
[1] H. Tataria, M. Shafi, M. Dohler, and S. Sun, “Six critical challenges for 6G wireless systems: A summary and some solutions,” IEEE Veh. Technol. Mag., vol. 17, no. 1, pp. 16–26, Mar. 2022, doi: 10.1109/MVT.2021.3136506.
[2] M. Adhikari, A. Hazra, V. G. Menon, B. K. Chaurasia, and S. Mumtaz, “A roadmap of next-generation wireless technology for 6G-enabled vehicular networks,” IEEE Internet Things Mag., vol. 4, no. 4, pp. 79–85, Dec. 2021, doi: 10.1109/IOTM.001.2100075.
[3] L. Cheng, W. Viriyasitavat, M. Boban, and H.-M. Tsai, “Comparison of radio frequency and visible light propagation channels for vehicular communications,” IEEE Access, vol. 6, pp. 2634–2644, 2018, doi: 10.1109/ACCESS.2017.2784620.
[4] B. Turan and S. Coleri, “Machine learning based channel modeling for vehicular visible light communication,” IEEE Trans. Veh. Technol., vol. 70, no. 10, pp. 9659–9672, Oct. 2021, doi: 10.1109/TVT.2021.3107835.
[5] A.-M. Caˇilean and M. Dimian, “Current challenges for visible light communications usage in vehicle applications: A survey,” IEEE Commun. Surveys Tuts., vol. 19, no. 4, pp. 2681–2703, 4th Quart. 2017, doi: 10.1109/COMST.2017.2706940.
[6] P. Sharda, G. S. Reddy, M. R. Bhatnagar, and Z. Ghassemlooy, “A comprehensive modeling of vehicle-to-vehicle based VLC system under practical considerations, an investigation of performance, and diversity property,” IEEE Trans. Commun., vol. 70, no. 5, pp. 3320–3332, May 2022, doi: 10.1109/TCOMM.2022.3158325.
[7] Y. Zhang, L. Zhao, G. Zheng, X. Chu, Z. Ding, and K.-C. Chen, “Resource allocation for open-loop ultra-reliable and low-latency uplink communications in vehicular networks,” IEEE Trans. Veh. Technol., vol. 70, no. 3, pp. 2590–2604, Mar. 2021, doi: 10.1109/TVT.2021.3061582.
[8] H. Yang, K. Zheng, L. Zhao, and L. Hanzo, “Twin-timescale radio resource management for ultra-reliable and low-latency vehicular networks,” IEEE Trans. Veh. Technol., vol. 69, no. 1, pp. 1023–1036, Jan. 2020, doi: 10.1109/TVT.2019.2954462.
[9] T. Nawaz, M. Seminara, S. Caputo, L. Mucchi, F. S. Cataliotti, and J. Catani, “IEEE 802.15.7-compliant ultra-low latency relaying vlc system for safety-critical ITS,” IEEE Trans. Veh. Technol., vol. 68, no. 12, pp. 12,040–12,051, Dec. 2019, doi: 10.1109/TVT.2019.2948041.
[10] T. Gou, C. Wang, and S. A. Jafar, “Aiming perfectly in the dark-blind interference alignment through staggered antenna switching,” IEEE Trans. Signal Process., vol. 59, no. 6, pp. 2734–2744, Jun. 2011, doi: 10.1109/TSP.2011.2129517.
[11] M. Morales-Céspedes, M. C. Paredes-Paredes, A. García Armada, and L. Vandendorpe, “Aligning the light without channel state information for visible light communications,” IEEE J. Sel. Areas Commun., vol. 36, no. 1, pp. 91–105, Jan. 2018, doi: 10.1109/JSAC.2017.2774518.
[12] M. Karbalayghareh, F. Miramirkhani, H. B. Eldeeb, R. C. Kizilirmak, S. M. Sait, and M. Uysal, “Channel modelling and performance limits of vehicular visible light communication systems,” IEEE Trans. Veh. Technol., vol. 69, no. 7, pp. 6891–6901, Jul. 2020, doi: 10.1109/TVT.2020.2993294.
[13] M. Morales-Céspedes, H. Haas, and A. G. Armada, “Optimization of the receiving orientation angle for zero-forcing precoding in VLC,” IEEE Commun. Lett., vol. 25, no. 3, pp. 921–925, Mar. 2021, doi: 10.1109/LCOMM.2020.3038027.
[14] A. Nuwanpriya, S.-W. Ho, and C. S. Chen, “Indoor MIMO visible light communications: Novel angle diversity receivers for mobile users,” IEEE J. Sel. Areas Commun., vol. 33, no. 9, pp. 1780–1792, Sep. 2015, doi: 10.1109/JSAC.2015.2432514.
[15] L. Deng and Y. Fan, “Analysis of channel correlation and channel capacity for indoor MIMO visible light communication systems,” Appl. Opt., vol. 59, no. 15, pp. 4672–4684, May 2020, doi: 10.1364/AO.386587.
Digital Object Identifier 10.1109/MVT.2022.3228389