LiFi as a Paradigm-Shifting 5G Technology

In this paper we argue that the optical spectrum could transform wireless network in a similar way and why it has transformed wired communication networks with the advent of fiberoptical communication. Peak transmission speeds with off-the-shelf light emitting diodes (LEDs) of 15.7 Gbps have recently been demonstrated. We will explain how these advances will be used to build full wireless networks which support user mobility. We discuss numerous misconceptions and use cases. Lastly, we illustrate the potential impact this technology may have on new and emerging industries.

Introduction

LiFi is a wireless communication technology that uses the infrared and visible light spectrum for high speed data communication. LiFi, first coined in [1] extends the concept of visible light communication (VLC) to achieve high speed, secure, bi-directional and fully networked wireless communications [2]. It is important to note that LiFi supports user mobility and multiuser access. The size of the infrared and visible light spectrum together is approximately 2,600 times the size of the entire radio frequency spectrum of 300 GHz (Figure 2). It is shown in [3] that the compound annual growth rate (CAGR) of wireless traffic has been 60% during the last 10 years. If this growth is sustained for the next 20 years, which is a reasonable assumption due to the advent of Internet-of-Things (IoT) xK-TV and machine type communication (MTC), this would mean a demand of 12,000 times the current bandwidth assuming the same spectrum efficiency. As an example, the industrial, scientific and medical (ISM) RF band in the 5.4 GHz region is about 500 MHz, and this is primarily used by wireless fidelity (WiFi). This bandwidth is already becoming saturated, which is one reason for the introduction of Wireless Gigabit Alliance (WiGig). WiGig uses the unlicensed spectrum between 57 GHz – 66 GHz, i.e., a maximum bandwidth of 9 GHz. In 20 years from now, the bandwidth demand for future wireless systems would however, be 12,000 × 500 MHz which results in a demand for 6 THz of bandwidth. The entire RF spectrum is only 0.3 THz. This means a 20 times shortfall compared to the entire RF spectrum. In comparison, the 6 THz of bandwidth is only 0.8% of the entire IR and visible light spectrum. One could argue that a more aggressive spatial reuse of frequency resources could be adopted to overcome this looming spectrum crunch. This approach has been used very successfully in the past and has led to the 'small cell concept'. In fact, it has been the major contributor towards the improvements of data rates as illustrated in figure 1. The cell sizes in cellular communication have dramatically shrunk. The cell radius in early 2G systems was 35 km, in 3G systems 5 km, in 4G systems 100 m, and in 5G probably about 25 m in order to reuse the available RF spectrum more efficiently and to achieve higher data densities.

However, further reductions in cell sizes are more difficult to achieve due to the high infrastructure cost for the backhaul and fronthaul data links which connect these distributed access points to the core network. Moreover, with a smaller cell size the likelihood of line-of-sight between an interfering base station and a user terminal increase. The resulting interference can significantly diminish data rates and may cause a major problem in cellular networks [4]. Therefore, WiFi access points have been mounted under the seats in stadia to use the human body as an attenuator for the RF signals and to avoid line-of-sight interference links. Clearly, this is not a viable solution for office and home deployments. For these reasons, it is conceivable that the contributions for the future mobile data traffic growth will stem from more spectrum rather than spatial reuse. In particular, the optical resources are very attractive as they are plentiful as shown in figure 2 and they are license-free.

Figure 2: The radio frequency (RF) spectrum is only a fraction of the entire electromagnetic spectrum. The visible light spectrum and the infrared (IR) spectrum are unregulated, and offer 780 THz of bandwidth

These resources can be used for data communication which is successfully demonstrated for decades in fiber-optic communication using lasers. With the widespread adoption of high brightness light emitting diodes (LEDs) an opportunity has arisen to use the visible light spectrum for pervasive wireless networking.

Traditionally, a VLC system has been conceived as a single point-to-point wireless communication link between a LED light source and a receiver which is equipped with a photo detection device such as a photo detector (PD). The achievable data rate depends on the digital modulation technology used as well as the lighting technology. The available lighting technologies are summarized in figure 3.

Figure 3: The maximum achievable data rates in LiFi depend on the technology of the actual light sources. Here we consider single blue-chip technology with phosphorous coating; red, green and blue (RGB) LEDs; Gallium Nitride (GaN) micro LEDs and laser-based lighting

Most commercial LEDs are composed of a blue high brightness LED with a phosphorous coating that converts blue light into yellow. When blue light and yellow light are combined, this turns into white light. This is the most cost-efficient way to produce white light today, but the phosphor color converting material slows down the frequency response, i.e., higher frequencies are heavily attenuated. Consequently, the bandwidth of this type of LED is merely in the region of 2 MHz.  With a blue filter at the receiver to remove the slow yellow components it, however, is possible to achieve data rates in the region of 1 Gbps with these devices. More advanced red, green and blue (RGB) LEDs enable data rates up to 5 Gbps as white light is produced by mixing the base colors instead of using a color converting chemical. Record transmission speeds with a single micro LED of 8 Gbps have been demonstrated [5], and it was shown that 100 Gbps are feasible with laser-based lighting [6].

The key advantages of a LiFi wireless networking layer are:
•    three orders of magnitude enhanced data densities [7];
•    unique properties to enhance physical layer security [8];
•    use in intrinsically safe environments such as petrochemical plants and oil platforms where RF is often banned;
•    with the advent of power-over-ethernet (PoE) and its use in lighting, there exists the opportunity to piggy-back on existing data network infrastructures for the required backhaul connections between the light sources with its integrated LiFi modem, and the Internet.

LiFi Networking

Figure 4 illustrates the concept of a LiFi attocell networking. The room is lit by several light fixtures. Each light is driven by a LiFi modem or a LiFi chip and, therefore, also serves as an optical base station or access point (AP). The optical base stations are connected to the core network by high speed backhaul connections. The light fixtures also have an integrated infrared detector to receive signals from the terminals. The high frequency flickers resulting from Mbps and Gbps data encoding are much higher than the refresh rate of a computer monitor, and hence these flickers are not visible to the occupants of the room. Power and data can be provided to each light fixture using a number of different techniques, including PoE and power-line communication (PLC) [9][10]. An optical uplink is implemented by using a transmitter on the user equipment (UE), often using an IR source (so it is invisible to the user). Each of these light fixtures, which at the same time act as wireless LiFi APs, create an extremely small cell, an optical attocell [11]. Because light is spatially confined, it is possible in LiFi to take the 'small cell concept' to a new level by creating ultra-small cells with radii less than 5 m while exploiting the huge additional unlicensed spectrum in the optical domain. The balance of light fixtures that contain APs and those that provide only illumination is determined by the requirement of the network, but potentially all light fixtures can contain APs. Compared to a single AP wireless hot-spot system, such cellular systems can cover a much larger area and allow multiple UEs to be connected simultaneously [12]. In cellular networks, dense spatial reuse of the wireless transmission resources is used to achieve very high data density - bits per second per square meter (bps/m2). Consequently, the links using the same channel in adjacent cells interfere with each other, which is known as co-channel interference (CCI) [13]. Figure 5 illustrates CCI in an optical attocell network.

Figure 4: LiFi attocell networking supports mobile terminals; bi-directional communication links; multiple mobile or fixed terminals connected to a single luminaire as well as a high speed backhaul

Figure 5: CCI occurs in the region where the same light spectrum of neighboring APs overlaps, and when these APs use the same modulation bandwidth for data encoding

The move from point to point links to full wireless networks based on light, poses several challenges. Within each cell, there can be several users and therefore multiple access schemes are required. The provision of an uplink can also require a different approach from the downlink. This is because low energy consumption is required in the portable device, and an uplink visible light source on the device is likely to be distracting to the user. Therefore, the use of the infrared spectrum seems most appropriate for the uplink. In addition, modulation techniques for a high-speed uplink have to be spectrum efficient and power efficient at the same time. Two recently developed modulation techniques that achieve this are enhanced unipolar OFDM (eU OFDM) [14], or spectral and energy efficient (SEE OFDM) [15].

Interference mitigation techniques are required to ensure within the region of strong CCI, a mobile station can also achieve high SINR, and this is a non-trivial problem which involves signal processing such as successive interference cancellation [23]. Alternative CCI mitigation techniques [16] include the use of intelligent resource schedulers. The main tasks of the 'resource scheduler' are to adaptively allocate signal power, frequency, time and wavelength resources. Typically, there are trade-offs between signaling overhead, computational complexity, user data rates, aggregate data rates and user fairness, and the optimum selection of respective CCI mitigation and resource scheduling techniques depend on actual use cases and system constraints [18][19]. Other functions of the central controller include achieving multi-user, and the handover process from cell to cell when terminals move. Handover plays an important role in LiFi networks. For example, the handover controller has to ensure that connectivity is maintained when users leave a room, or the premises. Therefore, there might be situations when there is no LiFi coverage. In these scenarios, to avoid loss of connectivity, we utilize the fact that LiFi is complementary to RF networks. To this end, there have been studies on hybrid LiFi/RF networks leading to three key-findings.

The three key findings on hybrid LiFi/RF networks studies:
•    LiFi networks will significantly improve services quality to mobile users,
•    service delivery can be uninterrupted, and
•    WiFi networks significantly benefit from LiFi networks. The latter is
     because well-designed load balancing will ensure that WiFi networks
     suffer less from inefficient traffic overheads caused by constant
     re-transmissions which happen when two or multiple terminals are
     in contention [20].

LiFi attocell networks have many advantages over incumbent technologies. Firstly, unlike omnidirectional RF antennas radiating signals in all directions, a LED light source typically radiates optical power directionally because of the way it is constructed. Therefore, the radiation of the visible light signals is naturally confined within a limited region. In contrast, RF mm-wave systems require complicated and expensive antenna beamforming techniques to achieve the same objective. Secondly, LiFi attocell networks can be implemented by modifying existing lighting systems. Any LiFi attocell network can provide extra wireless capacity without interference to RF networks that may already exist. LiFi attocell networks, therefore, have the potential to augment 5G cellular systems in a cost-effective manner [21].

A unique feature of LiFi is that it combines illumination and data communication by using the same device to transmit data and to provide lighting. Figure a depicts a simple room scenario with two lights. Figure b shows the resulting illuminance at desk level of 0.75 m. In the particular example, the lights are placed such that within the plane at desk height, 90% of the area achieves an illuminance of 400 lux based on a given illumination requirement. Figure c depicts the resulting signal-to-interference-plus-noise ratio (SINR). The region where the light cones overlap is subject to strong CCI, and the SINR drops significantly. It is interesting to note that the SINR can vary by about 30 dB within a few centimeters. This example also highlights that the peak SINR can be in region of 50 dB which is two to three orders of magnitude higher than the peak SINR in RF based wireless systems. The achievable data rate strongly depends on the location of the receiver and also on the field of view (FoV) of the receiver [22]. It should be noted that two lights in neighboring rooms which are separated by an opaque wall will cause no mutual CCI. This is fundamentally different from RF networks where radio signals propagate through walls and cause co-channel interference within a wide area. Because of this property of RF communications, it is difficult to achieve very high data densities.

Figure 6: A room of size 2.5 m × 5 m is equipped with two LiFi luminaires installed at 3 m height pointing vertically downwards. The LiFi luminaires are illustrated by two blue squares in subplot (a). Both luminaires use the same visible light spectrum to transmit independent information. Vertically upwards pointing receivers at 0.75 m desk height are assumed. The illuminance at desk height is illustrated in subplot (b). The resulting SINR assuming a receiver FoV of 45° is depicted in subplot (c)

The feature that light does not propagate through opaque objects can also be used to enhance data security (Figure 7). Our latest research has shown that physical layer security can be enhanced by a factor of 20 compared to existing WiFi.  

Figure 7: Light signals are blocked by walls. This feature significantly enhances security in LiFi networks compared with WiFi networks. In a room with no WiFi router installed (left picture) it is possible to receive WiFi signals from outside the room. In contrast, in a room with opaque walls it is not possible to receive LiFi signals from outside the room (right picture). Likewise, signals from inside the room do not penetrate to neighboring rooms which are separated by walls

LiFi Misconceptions

In the following we'll especially discuss five misconceptions about LiFi.

LiFi is an LoS technology

This perhaps is the greatest misconception. By using an orthogonal frequency division multiplexing (OFDM)-type intensity modulation (IM)/direct detection (DD) modulation scheme [24], the data rate scales with the achieved signal-to-noise-ratio (SNR). This means higher order digital modulation schemes can be used in conjunction with OFDM to harness the available channel capacity. By using adaptive modulation and coding (AMC) it is possible to transmit data at SNRs as low as -6 dB due to the use of forward error correction (FEC) coding. Figure 8 illustrates a video transmission to the laptop in the front over a distance of about 3 m where the LED light fixture is pointing against a white wall in the opposite direction to the location of the receiver. Therefore, there is no direct LoS component reaching the receiver at the front, but the video is successfully received. Obviously, if the wall would be dark, more light would be absorbed which would compromise the SNR at the receiver. If the SNR drops below the -6 dB threshold, an error-free communication link would not be possible. However, in low-light conditions single photon avalanche diodes may be used at the receiver which enhance the receiver sensitivity by at least an order of magnitude [26].

Figure 8: This illustration shows the operation of a LiFi link under strict non-line-of-sight (LoS) conditions (courtesy pureLiFi)

LiFi does not work in sunlight conditions

Sunlight constitutes a constant interfering signal outside the bandwidth used for data modulation. LiFi operates at frequencies typically greater than 1 MHz. Therefore, constant sunlight can be removed using electrical filters. An additional effect of sunlight is enhanced shot noise, which cannot easily be eliminated by electrical filters. In a study [27] the impact of shot noise was investigated qualitatively, and it was found that data rate is compromised by about 5%. Saturation can be avoided by using automatic gain control algorithms in combination with optical filters. In fact, we argue that sunlight is hugely beneficial as it enables solar cell based LiFi receivers where the solar cell acts as data receiver device, and at the same time harvests sunlight as energy [28].

Lights cannot be dimmed

There  are advanced modulation techniques such as eU-OFDM [14] which enable the operation of LiFi close to the turn-on voltage (ToV) of the LED which means that the lights can be operated at very low light output levels while maintaining high data rates.

The lights flicker

The lowest frequency at which the lights are modulated is in the region of 1 MHz. The refresh rate of a computer screen is about 100 Hz. This means the flicker-rate of a LiFi light bulb is 10,000 higher than that of a computer screen. Therefore, there is no perceived flicker.

This is for downlink only

A key advantage is that LiFi can be combined with LED illumination. This, however, does not mean that both functions always have to be used together. Both functions can easily be separated (see the comment on dimming). As a result, LiFi can also be very effectively used for uplink communication where lighting is not required. The infrared spectrum, therefore, lends itself perfectly for the uplink. We have conducted an experiment where we sent data at a speed of 1.1 Gbps over a distance of 10 m with an LED of only 4.5 mW optical output power.

LiFi Applications

LiFi applications are manifold. As shown in figure 9, streetlights could play a major role in future smart cities. They could provide gigabit bi-directional wireless connectivity. Interestingly, with LiFi this would also be possible during daytime. As illustrated in figure, LiFi can also unlock smart transport systems which are part of our fully connected smart cities. Because cars typically use LEDs or lasers in headlights and taillights, these lights can be used for gigabit inter-car communication, and data communication to street furniture such as traffic lights and street lights.

Unlike RF communication systems, LiFi will also work underwater as shown in figure. This will allow new ways to connect remote operated vehicles, and it also will allow divers to communicate with each other. This will enhance safety in difficult underwater missions. Transmission distances of up to 100 m have been demonstrated. New detector technology such as single photon avalanche detectors (SPADs) are currently being investigated to achieve much higher distances.

Figure 9: Streetlights could form the backbone of future 5G networks in cities

Figure 10: LiFi will work underwater, and the lights that are used for illumination in remote operated vehicle and underwater drones can be used to exchange information among these autonomous underwater machines. It will be possible to create mesh networks to send information over long distances under water to create sensor networks for environmental monitoring

Figure 11: Cars can communicate with each other to avoid accidents, and connectivity between cars will also aid the trend of 'driverless' cars. Moreover, cars can communicate to street furniture such as traffic lights and streetlights to create smart transport systems in our crowded cities

An additional important application area is the internet of things (IoT). The IoT can be classified in the industrial IoT, and the general IoT. The latter may connect our future appliances such as our toaster, microwaves, freezers, ovens and fridges to the internet. In fact, LiFi will enable the LED status lights of these home appliances to connect them to the Internet via the domestic lighting system. This connectivity will act as the 'nervous system' for our future things. Assume embedded microprocessors, sensor, memory and machine learning algorithms, LiFi will provide the high bandwidth, low latency connectivity to realize meaningful artificial intelligence. For example, by using predictive maintenance it will be possible to establish if an item is liable to break in the near future. The item could then order itself or as specified by the user (before it breaks) from the Internet automatically, avoiding inconvenient disruptions, and saving us time. Another important application of LiFi is indoor positioning and navigation. Typically, there are many lights in an indoor environment. Light is confined to a small area. This feature can be exploited to readily establish the position of people and assets within a few meters.

Market Disruption Potential

LiFi is a disruptive technology that is poised to impact a large number of industries. LiFi is a fundamental 5G technology. It can unlock the IoT, drive Industry 4.0 applications, light-as-a-service (LaaS) in the lighting industry, enable new intelligent transport systems, enhance road safety when there are more and more driver-less cars, create new cyber-secure wireless networks, enable new ways of health monitoring in aging societies, offer new solutions to close the digital divide and enable very high-speed wireless connectivity in future data centers.

LiFi will have a catalytic effect for the merger of two major industries:
•    the wireless communications industry and
•    the lighting industry as illustrated in figure 12

Figure 12: LiFi has the potential to act as the catalyst of a process that ultimately leads to a merger of the wireless communications industry and the lighting industry

Figure 12 demonstrates the vision of how LiFi could lead to a merger of the wireless communications industry and the lighting industry. In the lighting industry, LiFi provides a means to diversify and to develop new applications and this will propel the trend of light-as-a-service (LaaS). This will pull the lighting industry into the markets of the wireless communications industry. As a result, new business models will be created in the lighting industry which are needed as the life-time of an LED light bulb is 20 years and more. The wireless communication industry requires unprecedented data rates and orders of magnitude higher data densities due to new services in 5G such as augment reality and virtual reality as well as mobile TV. In addition, wireless networks will need to connect billions of internet-of-things (IoT) devices. This will accelerate the radio frequency spectrum crunch, and 'LiFi' will act as a 'pressure valve' which means that there will be a market 'push' to develop wireless communications equipment based on light, a market that has classically been served by the lighting industry

In 25 years from now, we moot that the LED lightbulb will serve thousands of applications and will be an integral part of the emerging smart cities, smart homes and the IoT. LaaS will be a dominating theme in the lighting industry, which will drive the required new business models when LED lamps last 20 years or more. LaaS in combination with LiFi will, therefore, provide a business model driven 'pull' for the lighting industry to enter what has traditionally been a wireless communications market. In the wireless industry, LiFi has the potential to create a paradigm shift by moving from cm-wave communication to nm-wave communication. It is, therefore, conceivable that the wireless industry and the lighting industry will merge into one. An important prerequisite for the large-scale adoption of LiFi technology is the availability of standards. In this context, efforts have started in IEEE 802.15.7, IEEE 802.11 as well as ITU-R to standardize LiFi technology. Notably, there is now a Task Group on Light Communication within 802.11bb.

Conclusion

The visible light spectrum and infrared spectrum together, offer 2600 times more bandwidth than the entire RF spectrum. LiFi harnesses the abundance of bandwidth to achieve new wireless networks which augment existing RF-based wireless networks. These networks increasingly suffer from bandwidth shortages. With current commercially available optical devices it is possible to achieve multi-gigabit bi-directional data links. LiFi integrates these links into a full wireless network which is augmented with functions such as multiuser access, handover and CCI mitigation. This paper has attempted to clarify a number of misconceptions about LiFi. A few selected use cases have been discussed. It has been shown that LiFi has the potential to lead to a merger of the lighting and the wireless communication industries. Therefore, LiFi has become a reality and this technology is here to stay for a long time.

Acknowledgements:

Professor Harald Haas gratefully acknowledges the support by the Engineering and Physical Research Council (EPSRC) under Grant EP/R007101/1. He also acknowledges the financial support by the Wolfson Foundation and the Royal Society.

References:
[1]    H. Haas, "Wireless data from every light bulb." TED Global, Aug-2011.
[2]    H. Haas, L. Yin, Y. Wang, and C. Chen, "What is LiFi?," IEEE J. Light. Technol., vol. 34, no. 6, pp. 1533–1544, Mar. 2016.
[3]    P. J. Winzer and D. T. Neilson, "From scaling disparities to integrated parallelism: a decathlon for a decade," IEEE J. Light. Technol., vol. 35, no. 5, pp. 1099–1115, Mar. 2017.
[4]    F. Boccardi, R. W. Heath, A. Lozano, T. L. Marzetta, and P. Popovski, "Five disruptive technology directions for 5G," IEEE Commun. Mag., vol. 52, no. 2, pp. 74–80, Feb. 2014.
[5]    M. S. Islim et al., "Towards 10 Gb/s orthogonal frequency division multiplexing-based visible light communication using a GaN violet micro-LED," Photon. Res., vol. 5, no. 2, pp. A35--A43, 2017.
[6]    D. Tsonev, S. Videv, and H. Haas, "Towards a 100 Gb/s visible light wireless access network," Opt. Express, vol. 23, no. 2, pp. 1627–1637, 2015.
[7]    I. Stefan, H. Burchardt, and H. Haas, "Area spectral efficiency performance comparison between VLC and RF femtocell networks," in IEEE International Conference on Communications (ICC), 2013, pp. 3825–3829.
[8]    Y. Liang and H. Haas, "Physical-Layer Security in Multiuser Visible Light Communication Networks," IEEE J. Sel. Areas Commun., 2017.
[9]    W. Ni, R. P. Liu, B. Collings, and X. Wang, "Indoor Cooperative Small Cells over Ethernet," IEEE Commun. Mag., vol. 51, no. 9, pp. 100–107, 2013.
[10]    A. Papaioannou and F. N. Pavlidou, "Evaluation of Power Line Communication Equipment in Home Networks," IEEE Syst. J., vol. 3, no. 3, pp. 288–294, 2009.
[11]    H. Haas, "High-speed wireless networking using visible light," SPIE Newsroom, 2013.
[12]    V. H. MacDonald, "The cellular concept," Bell Syst. Tech. J., vol. 58, no. 1, pp. 15–43, Jan. 1979.
[13]    A. Goldsmith, Wireless Communications. Cambridge University Press, 2005.
[14]    D. Tsonev, S. Videv, and H. Haas, "Unlocking spectral efficiency in intensity modulation and direct detection systems," IEEE J. Sel. Areas Commun., vol. PP, no. 99, p. 1, 2015.
[15]    H. Elgala and T. D. C. Little, "SEE-OFDM: Spectral and energy efficient OFDM for optical IM/DD systems," in 2014 IEEE 25th Annual International Symposium on Personal, Indoor, and Mobile Radio Communication (PIMRC), 2014, pp. 851–855.
[16]    H. Ma, L. Lampe, and S. Hranilovic, "Robust MMSE linear precoding for visible light communication broadcasting systems," in 2013 IEEE Globecom Workshops (GC Wkshps), 2013, pp. 1081–1086.
[17]    A. Tzanakaki et al., "Wireless-Optical Network Convergence: Enabling the 5G Architecture to Support Operational and End-User Services," IEEE Commun. Mag., vol. 55, no. 10, pp. 184–192, 2017.
[18]    B. Ghimire and H. Haas, "Self-organising interference coordination in optical wireless networks," EURASIP J. Wirel. Commun. Netw., vol. 1, no. 131, Apr. 2012.
[19]    C. Chen, V. S., D. Tsonev, and H. Haas, "Fractional frequency reuse in DCO-OFDM-based optical attocell networks," J. Light. Technol., vol. 33, no. 19, pp. 3986–4000, Oct. 2015.
[20]    Y. Wang, S. Videv, and H. Haas, "Dynamic load balancing with handover in hybrid Li-Fi and Wi-Fi networks," in Proc. IEEE 25th International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC), 2014, pp. 548–552.
[21]    M. Ayyash et al., "Coexistence of WiFi and LiFi toward 5G: concepts, opportunities, and challenges," IEEE Commun. Mag., vol. 54, no. 2, pp. 64–71, Feb. 2016.
[22]    I. Stefan and H. Haas, "Analysis of Optimal Placement of LED Arrays for Visible Light Communication," in Proc. of IEEE Vehicular Technology Conference (VTC Spring), 2013, p. 5 pages.
[23]    P. Patel and J. Holtzmann, "Analysis of a Simple Successive Interference Cancellation Scheme in DS/CDMA System," IEEE J. Sel. Areas Commun., vol. 12, no. 5, pp. 796–807, 1994.
[24]    M. Z. Afgani, H. Haas, H. Elgala, and D. Knipp, "Visible light communication using OFDM," in Proc. IEEE 2nd Int. Conf. Testbeds Res. Infrastructures Develop. Netw. Communities, 2006, pp. 129–134.
[25]    K. D. Langer and J. Vucic, "Optical Wireless Indoor Networks: Recent Implementation Efforts," in Proc. European Conference on Optical Communication (ECOC), 2010, p. 6 pages.
[26]    Y. Li, M. Safari, R. Henderson, and H. Haas, "Optical OFDM with single-photon avalanche diode," IEEE Photonics Technol. Lett., vol. 27, no. 9, pp. 943–946, 2015.
[27]    M. S. Islim, M. Safari, S. Videv, and H. Haas, "A Proof-of-Concept of Outdoor Visible Light Communications in the presence of Sunlight," in LED professional Symposium - Expo 2016, 2016.
[28]    Z. Wang, D. Tsonev, S. Videv, and H. Haas, "On the design of a solar-panel receiver for optical wireless communications with simultaneous energy harvesting," IEEE J. Sel. Areas Commun., vol. 33, no. 8, pp. 1612–1623, Aug. 2015.