We explore the design of an LED-based communication system comprising two free space optical links: one narrow-beam (primary) link for bulk data transmission and one wide-beam (beacon) link for alignment and support of the narrow-beam link. Such a system combines the high throughput of a highly directional link with the robust insensitivity to pointing errors of a wider-beam link. We develop a modeling framework for this dual-link configuration and then use this framework to explore system tradeoffs in power, range, and achievable rates. The proposed design presents a low-cost, compact, robust means of communication at short- to medium-ranges, and calculations show that data rates on the order of Mb/s are achievable at hundreds of meters with only a few LEDs.
© 2014 Optical Society of America
Since optical wireless (OW) communications was first proposed decades ago , there has been much exploration of its potential advantages as an alternative to traditional radio-frequency (RF) wireless technology. Unlike RF communications, OW uses spectrum that is relatively unregulated and offers large modulation bandwidths for communication . Communication at these optical wavelengths can also provide high levels of security against eavesdropping and jamming, as optical links are often highly directional in nature and are typically unable to penetrate walls and other obstacles . These advantages have motivated the study of OW for applications ranging from long range laser-based communications [4–7] to indoor local-area networks [8–12].
However, OW also presents its own constraints and challenges. The conversion at the receiver of intensity-modulated light to an electrical signal is fundamentally different than the conversion of an RF signal and can require a higher concentration of power incident on the receiver [2, 3]. Thus, for ranges beyond a few meters, it is often necessary for optical links to focus and precisely direct the transmitted energy onto the receiver in order to achieve desired data rates and ranges. For example, a mechanically pointed laser-based link operating at a range of 1 km can require pointing precision on the order of 10μrad to 1–2 mrad [7, 13], which allows for very high data rates (e.g., on the order of 1 Gb/s) but can require bulky, specialized mechanical platforms. Such precise alignment can be also be a challenge to maintain, especially in mobile networks . To address these alignment challenges, long-range links are often supported by laser-based beacons that are somewhat wider than the primary data laser have also been used to transmit positioning information and alignment feedback, but these beams are still relatively narrow (on the order of one milliradian to tens of milliradians) and assume knowledge of coarse positioning information , . Independent RF links are often used in conjunction with laser-based FSO links, to either transmit alignment information between nodes and decrease the occurrence of complete link nonavailability or form hybrid RF/FSO networks [5, 12, 15–20].
Light-emitting diode (LED) technology is maturing into the low-cost, energy-efficient lighting source of the future [21,22], and this maturation presents an opportunity to create ubiquitous point-to-point OW links in a different regime from that of laser-based systems. Specifically, the low power consumption, compactness, low cost, reliability, and diffuse emission patterns of LEDs make them attractive candidates for use in short- to medium-range (tens to hundreds of meters) data links supporting Mb/s rates. They could be especially useful in mobile scenarios where very precise alignment can be prohibitively difficult, including some vehicle-to-vehicle or robot-to-robot communication applications. Communication in these applications has often utilized RF technology, which can sometimes suffer from external interference or generate unwanted interference in sensitive environments. LED-based communication, however, could maintain near-complete immunity to jamming interference and has much less potential to interfere with nearby electronic devices. Such technology thus has the potential to provide a wireless, robust, low-cost, secure means of communication in settings in which radio communication may cause or suffer from unwanted interference. This may include some tactical military environments and healthcare settings with sensitive electronics.
With this application space in mind, we explore the use of a directed OW communication system in which a robust, lower-data-rate, wide-beam LED-based link is used to provide acquisition and alignment support for a more tightly focused, higher-data-rate LED-based link. For both links, the beamwidths examined here are on the order of tens of degrees, which is orders of magnitude wider than what is often considered in FSO systems (10μrad to 1 mrad). The relaxed alignment constraints of the proposed system make it well suited for mobile networks. To the best of the authors’ knowledge, this is the first discussion of a system that simultaneously exploits the robustness of a very wide-beam LED link and the relatively high data throughput of a more narrow-beam link.
We begin in Section 2 by describing a link model for intensity-modulation direct-detection (IM/DD) LED-based systems. In Section 3, we apply this model to investigate achievable ranges and robustness using a wide-beam link and then, based on this investigation, develop a framework for the design of a dual-link (wide- and narrow-beam) system. We illustrate this framework with calculations of achievable ranges and rates assuming one or a few LEDs, although the analysis is more generally applicable. Finally, we provide some concluding remarks in Section 4.
2. Link model
We consider a link model of a line-of-sight (LOS) IM/DD OW system that employs on-off keying (OOK) modulation. We assume that the transmitter has one or a few LEDs that emit light into a hemisphere and that the pattern of emission can be described by an irradiance function Is(d, ϕ) [W/m2] given by  asFig. 1. The case of ϕ = 0 corresponds to a transmitter that is perfectly pointed at the receiver, and thus we will sometimes refer to ϕ as the “pointing error.” The average transmitted optical power is P [W], and m is a parameter defining the beamwidth of emission. The half-power half-beamwidth Φ1/2, which we will refer to as the “beamwidth,” is related to m by
For a receiver placed at a location defined by (d, ϕ), the received optical signal power PRx is given byFig. 1). The case of ψ = 0 corresponds to a receiver that is perfectly pointed at the transmitter. For a receiver that is composed of a photodiode of active area A, an optical filter described by the parameter Ts(ψ), and an optical concentrator of gain g(ψ), the effective area is 24] 23]. The case of no optical concentrator corresponds to a case of a concentrator with n = 1 (free space) and Ψc = 90°, yielding a gain of g = 1.
In OW systems, there are many potential sources of noise, including thermal noise in the receiver, artificial lighting [25, 26], and shot noise from the ambient sunlight . Often in OW systems, and especially for outdoor systems, the dominant noise source is shot noise from isotropic ambient light [3, 23, 26–28]. To reduce the ambient optical power Pn [W] that is received by the photodiode, an optical passband filter can be placed on the receiver. In calculating the effect of this filter on the noise level, we model it as a an ideal “boxcar” passband filter of spectral width Δλ [nm]. The filter has a transmittance Tn within the passband and zero outside the passband. A practical filter may have an angularly depedendent transmittance, but can be approximately modeled as a “boxcar” filter of effective passband width Δλ. We also assume that the ambient background noise incident on the receiver is “white” (constant within the pass-band), and define its spectral irradiance (power per unit photodetector area per unit spectrum) as pbg [W/nm-cm2]. With an ideal optical concentrator of index of refraction n, the ambient optical power incident on the photodiode is 
This ambient light creates shot noise in the receiver, which is typically modeled as introducing zero-mean additive white Gaussian noise (AWGN) to the received photocurrent, where the variance [A2] of the AWGN can be approximated by 
We use this noise model to calculate the signal-to-noise ratio (SNR) at the receiver, which we define as23]. We can relate the bit rate B, ambient shot noise level, average transmitted power, range, beamwidth, and BER. Combining Eq. (8)–(10) and solving for B yields the rate Eqs. (1) and (3) into Eq. (11), yielding
3. Design of wide beam/narrow-beam dual link system
3.1. Defining the role of the beacon link
Using the link model of Section 2, we explore the use of a wide-beamwidth LED-based link acting as a support link for a more focused, narrow-beam link. We will refer to the wide-beam link as the beacon link, and the narrow-beam link as the primary link. Throughout this discussion, subscripts b and p will be used to denote parameters relevant to the beacon link and primary link, respectively. For instance, we define Ψc,b as the concentrator field of view for the beacon link receiver and Ψc,p as that for the primary link receiver. The pointing angles of the beacon and primary transmitters are ϕb and ϕp, respectively, and the pointing angles of the beacon and primary receivers are ψb and ψp, respectively. To avoid interference between the two links, there is a need to ensure orthogonality between them; this could be achieved, for instance, by using LEDs of different wavelengths for the two links or time-division multiplexing their communication.
The primary link has a more focused beam than the beacon link and is expected to support a much higher data throughput than the beacon link. Operating such a relatively directional link, however, can introduce alignment challenges, especially in mobile scenarios. To address this, we propose the joint use of the supporting beacon link. The beacon link need not provide a high data rate; rather, its purpose is to provide low-data-rate connectivity for a wide range of transmitter pointing angles ϕb. This low rate connectivity could be used, for example, to provide positioning and alignment information for the primary link. There are many different ways this supporting link could help align the primary link; among the demonstrated uses of supporting links in FSO systems have been the transmission of GPS coordinates, inertial orientation information, and received signal strength (RSS) . Regardless of the specific role chosen for the beacon link, the beamwidths we examine for both links are on the order of tens of degrees, which significantly relaxes alignment constraints relative to that of many FSO systems. By utilizing both links, the dual-link system exploits the robustness of the beacon link while maintaining the high throughput of a relatively focused primary link. This robustness makes it suitable for LED-based outdoor mobile applications, a regime that has been studied significantly less than the indoor local area network application space [8, 10, 27, 29, 30].
The beacon link provides robustness by virtue of its relatively large beamwidth Φ1/2,b, which relaxes the beacon pointing demands. In designing the exact beamwidth of the beacon transmitter, there is a tradeoff between this robustness in pointing and the transmitter-to-receiver distances (d) that allow for connectivity; narrower beams can allow for longer-distance links but demand that the beacon transmitter be pointed with relative precision, whereas links with wider beams are more limited in their range but allow for more relaxed pointing demands.
We approach the design of the beacon link beamwidth Φ1/2,b by specifying a constraint on the pointing precision of the beacon transmitter. Specifically, we demand that the greatest pointing error allowed is |ϕb| = θa; in some sense, this defines an “angular range” of operation for the beacon link. In addition, we demand that for each ϕb within this permitted angular range (−θa ≤ ϕb ≤ θa), the beacon link supports a minimum data rate Bb0 (i.e., Bb ≥ Bb0). Note that this minimum rate is achievable at a different range d for each of the angles ϕb within this angular span.
The shortest of these distances d corresponds to |ϕb| = θa, the worst case of pointing within the stated constraints. We design the beacon link beamwidth Φ1/2,b to maximize this worst case range, because we are interested in optimizing the robustness of the beacon link over a wide range of pointing angles ϕb, rather than optimizing the performance of the link for cases of perfect pointing (ϕb = 0). To do this, we set ϕb = θa and differentiate Eq. (12) with respect to m. The parameter m defines the beamwidth via Eq. (2). The optimal m that results isEq. (12) yields the maximized range for this worst case of pointing, and we define this range as d0.
With this optimal beamwidth, beacon connectivity (Bb ≥ Bb0) is guaranteed to any receiver that lies d0 or less away from the transmitter, within the angular range −θa ≤ ϕb ≤ θa. Note that connectivity at ranges greater than d0 can be established for |ϕb| < θa, as well as for ranges less than d0 for |ϕb| > θa. A diagram that illustrates the geometry of the angular range |ϕb| ≤ θa and distance d0 is shown in Fig. 2. In practice, a single node can employ several beacons to “cover” a wider range of azimuthal and/or elevation angle, building on angle-diversity schemes that have been explored [31, 32]. However, the analysis in this work will focus on the use of a single beacon per node.
In general, the value of d0 depends on many parameters [see Eq. (12)], including the required beacon rate Bb0; very low values of Bb0 may be attainable at long distances, whereas higher rates may correspond to more limited ranges. The value of Bb0 itself depends on the desired used of the beacon link. Using the beacon link for acquisition and feedback control, for example, may require Bb0 ≈ 1 kb/s. Other uses of the beacon beam, such as allowing a receiver node to detect the presence of a beacon and perhaps calculate its bearing, might require lower rates. However, while the value of d0 depends on Bb0, ψb, and many other parameters, the optimal beamwidth Φ1/2,b depends only on the maximum allowed pointing error θa.
3.2. Exploring reasonable beacon rates and ranges
To calculate reasonable ranges and rates, we can use Eq. (11) to plot the beacon link rate as a function of the receiver position relative to the LED transmitter. Figure 3(a) shows a contour plot of the logarithm of the rates Bb over space, assuming that the receiver is pointed perfectly at the beacon transmitter (i.e., ψb = 0). Here, we choose to assume that the maximum allowed pointing error for the beacon is θa = 45°, and the beamwidth is optimized according to Eqs. (13) and (14) for this θa. The beacon transmitter is located at (X,Y) = (0,0) and is pointed in the positive Y-direction. In these calculations, we assume that the link uses a single high-power LED (beacon transmitting power Pb = 0.3 W) in bright daytime skylight noise (pbg = 5.8μW/nm/cm2 ). We also assume the receiver is composed of a colored glass filter of passband width Δλb = 100 nm and Ts,b = Tn,b = 0.8, a silicon p-i-n photodiode of responsivity R = 0.6 A/W and active area Ab = 1 cm2, and a glass optical concentrator (n = 1.5). Figure 3(b) assumes identical parameters, except that here the receiver is assumed to be poorly aligned. Specifically, it is misaligned by an amount equal to the transmitter maximum pointing error (ψb = θa = 45°). In both figures, we have a chosen receiver (and concentrator) field of view equal to the transmitter maximum pointing error (Ψc,b = θa = 45°). In practice, field of view varies among receivers, and there is no absolutely optimal field of view; rather, there is a tradeoff between field of view and gain, as seen in Eq. (6).
For the purposes of acquisition and feedback control, assuming a minimum beacon rate of Bb0 = 1 kb/s is reasonable. The calculations in Fig. 3(a) show that for an aligned receiver (ψb = 0), this required rate is achievable at d0 ≈ 85 m. If both the the transmitter and receiver are pointed perfectly (i.e., the receiver lies along the line X = 0, where ϕb = 0), then Bb = 1 kb/s is achievable at d ≈ 133 m. In the case of poor receiver alignment (ψb = θa = 45°), shown in Fig. 3(b), d0 is roughly 71 m.
In general, the sensitivity of d0 to the receiver pointing angle ψb depends on the optical concentrator gain [gb(ψb)], optical filter [Ts,b(ψb)], and a geometrical factor cos(ψb) [see Eqs. (5) and (12)]. Specifically, d0 is proportional to the square root of these factors. In the calculations presented in Fig. 3, the concentrator gain g is considered constant within its field of view defined by ψb < Ψc,b = θa. We also assume that Ts,b(ψb) is invariant in ψb for the beacon link, which is consistent with the behavior of an absorptive colored filter. Thus, in these calculations, the only dependence of d0 on the receiver misalignment ψb is the geometrical factor (cosψb)1/2. For the two receiver alignments examined here, [cos(ψb)]1/2 = 1 for the well-aligned receiver [Fig. 3(a)], and [cos(ψb)]1/2 ≈ 0.84 for the poorly aligned case [Fig. 3(b)]. Thus the ratio of the values of d0 in Fig. 3(a) and Fig. 3(b) is (71 m)/(85 m) ≈ 0.84.
3.3. Jointly designing the beacon and primary link
In designing a system that utilizes a beacon link to support a more focused link, we require that both links achieve the same range. Although the primary link may achieve useful data rates beyond the range at which the beacon link can achieve Bb = Bb0, we assume use of the primary link is contingent on successful operation of the beacon link. To meet this requirement of joint operation, it is necessary to consider the design space of the two links together. Figure 4 illustrates a representative example of this joint design space, where Fig. 4(a) describes the beacon link and Fig. 4(b) describes the primary link. The parameters assumed are the same as those of Fig. 3, except for θa and the beacon power Pb, parameters that are varied in Fig. 4(a).
Figure 4(a) defines a pair of curves for d0 as a function of beacon power Pb, one for ψb = 0 (well-aligned receiver, greater d0) and one for ψb = θa (misaligned receiver, shorter d0); this pair of curves is presented for three values of 2θa. Thus, for a given power Pb, θa, and receiver alignment ψb, the plot defines a range d0. This is the distance from the transmitter at which a data rate of Bb = 1 kb/s can be guaranteed within the angular range −θa ≤ ϕb ≤ θa. Taken alone, Fig. 4(a) is a design space only for the beacon link.
The ranges in Fig. 4(a) are strongly dependent on 2θa, but relatively weakly dependent on the receiver alignment. At all three ranges of 2θa, the ψb = 0 (well-aligned receiver) case corresponds to only a slightly greater range d0 than the poorly aligned case of ψb = θa. This weak dependence on ψb is a consequence of the choice of an incident-angle-insensitive filter and concentrator at the beacon receiver, as discussed at the end of the previous subsection. Note that this assumed misalignment θa changes for each value of 2θa examined; for 2θa = 40°, the misalignment considered is only ψb = 20°. Thus for this narrowest allowed angular range examined, the separation between the curves is small compared to that of the other two pairs.
For the beacon parameters chosen in Fig. 4(a), and for the calculated “worst-case” ranges d0, we next examine the data rates for the primary link with the assumption that its alignment is established and maintained by exploiting a beacon link of minimum data rate Bb = 1 kb/s. Thus we assume precise pointing for the primary link (ϕp = ψp = 0), even though for the primary beamwidths we examine (10° < Φ1/2,p < 20°), the primary link is not nearly as sensitive to pointing errors as typical long-range laser-based systems. To calculate the primary link data rate Bp, consider the ratio of Bp to Bb using Eq. (11). This yields
Figure 4(b) plots Bp as a function of Pp/Pb. For each of the values of θa examined in Fig. 4(a), Fig. 4(b) plots a pair of curves of primary-link data rates corresponding to two primary-link beamwidths (Φ1/2,p = 10° and Φ1/2,p = 20°), where rates corresponding to intermediate beamwidths lie between the paired curves. A common color-coding scheme is applied to Fig. 4(a) and Fig. 4(b), so that, for example, the two blue solid-line curves in Fig. 4(b) correspond to the case of 2θa = 90° in Fig. 4(a).
To calculate reasonable values of Bp, we assume different parameters for the primary link from those of the beacon link, including a smaller detector suited for higher modulation rates (Ap = 1 mm2 vs. Ab = 1 cm2) and a narrower bandpass filter (Δλp = 30 nm vs. Δλb = 100 nm) that can more effectively filter ambient noise. The other parameters in Eq. (15) assume values determined by Fig. 4(a), as the two plots are linked. For example, the transmitter pointing angle ψb and receiver field-of-view Ψb,c are dictated by the value of 2θa chosen in Fig. 4(a) and the previous assumptions that ψb = θa and Ψb,c = θa. The beamwidth parameter mb is determined by θa and Eq. (13). The beacon receiver is assumed to be either perfectly aligned (ψb = 0) or misaligned (ψb = θa) depending on the choice assumed in Fig. 4(a). We also assume that Ts,b(ψb) = Ts,p(ψb) = 0.8 and Ψp,c = 5°.
As a design example, we see that a beacon transmitter of 2θa = 90° with a range d0 = 117 m can be achieved at a power Pb = 0.57 W (roughly 1–2 high-power LEDs) for a misaligned receiver (ψb = θa = 45°). At this point in the design space, and at this range d0, Fig. 4(b) shows that a primary link of beamwidth Φ1/2,p = 10° using 0.24 times the beacon transmitter power (Pp/Pb = 0.24, Pp = 0.14W) can achieve a data rate of about 1 Mb/s. Note the sensitivity of the data rate to beamwidth, as increasing Φ1/2,p to 20° drops Bp to about 4.5 kb/s. To instead increase the primary-link data rate Bp by a factor kp, one could increase the power Pp by a factor of [see Eq. (11)]. For example, to achieve 10 Mb/s, one could boost the primary-link power such that Pp/Pb increases by a factor of [(10 Mb/s)/(1 Mb/s)]1/2, so that Pp/Pb = 0.77 and Pp = 0.44 W.
Maintaining the primary-link bit rate (Bp = 1 Mb/s) but instead extending the range (d0) of the dual-link system from 117 m to 500 m would require adjustments to both the beacon and primary links. At a beacon power of Pb = 0.57 W, a range of d0 = 500 m could be achieved by narrowing 2θa from 90° to 40°, as seen in Fig. 4(a). This adjustment would demand greater pointing precision for the beacon transmitter and receiver. Alternatively, this greater range could be reached by maintaining 2θa = 90° and increasing the power Pb by a factor of [(500m)/(117m)]2, as computed from Eq. (11). This power increase would require Pb = 10.4 W, a significant increase in the number of necessary LEDs. For reference, in the visible regime this might be on the order of two car headlights in terms of perceived brightness.
To extend the range to 500 m while maintaining the same data rate Bp = 1 Mb/s, the primary link would also have to be adjusted. One way to extend the primary-link range is to similarly increase the primary link power by a factor of [(500 m)/(117 m)]2. An alternative is to narrow the beamwidth Φ1/2,p [and thus increase the corresponding mp, defined by Eq. (2), according to Eq. (12)]. Specifically, adjusting the beamwidth from Φ1/2,p = 10° (mp = 45.28) to a narrower Φ′1/2,p (and larger m′p) requires following the relation , where kb = (500 m)/(117 m) in this example. Thus the beamwidth would be narrowed to Φ′1/2,p = 3 (m′p = 478.77) to support a rate of 1 Mb/s at a range of 500m.
We have demonstrated how Fig. 4 can be used to find reasonable ranges and rates in a dual-link system given desired power levels, beamwidths, and receiver alignments. The joint consideration of two links, primary and beacon, allows for specialization in the design of each link. Because the beacon link is to be robust, its transmission beam can be wider, its optical concentrator on the receiver has lower gain and a wider field of view, its optical bandpass filter is wider but incident-angle insensitive, and its detector area can be larger (to boost signal strength) due to lower data rates. The primary link is assumed to be more precisely aligned than the beacon link, even though its pointing demands are relaxed considerably relative to those of many laser-based systems. As the more focused, higher-throughput link, its receiver is designed to have a narrower-FOV/high-gain optical concentrator, a narrower interference-based bandpass filter (for enhanced noise rejection), and a smaller detector compatible with higher data rates. The use of these two complementary links can provide an all-optical LED-based system that is low power, compact, and robust to pointing and tracking error. This robustness may make this system a suitable adjunct to RF technology in short- to medium- range mobile networks.
We have presented a framework for the design of an all-optical LED-based dual-link system and have illustrated this framework with achievable range and rate predictions based on the practical parameter assumptions. The calculations show that in the presence of bright skylight it is possible to achieve a data rate of 1 Mb/s at over a hundred meters with relatively low transmitter power (e.g., < 1W optical power) and excellent robustness. This dual-link approach offers the potential for a robust, portable, compact system that can provide reasonable rates with a small number of LEDs, making it a possible adjunct or alternative to RF technology in mobile applications. The framework presented here could easily be expanded to explore optical systems that exploit many channels occupying different wavelengths over the visible and infrared domains. Future work may also explore the proposed dual-link system as the basis for multi-user networks.
Work at the University of Maryland was supported by a grant from the U.S. Army Research Office (ARO). The third author acknowledges support by ARO under grant number W911NF-13-1-0003.
References and links
1. F.R. Gfeller and U. Bapst, “Wireless in-house data communication via diffuse infrared radiation,” in Proceedings of the IEEE (IEEE, 1979), pp. 1474–1486. [CrossRef]
2. D.K. Borah, A.C. Boucouvalas, C.C. Davis, S. Hranilovic, and K. Yiannopoulos, “A review of communication-oriented optical wireless systems,” EURASIP Journal on Wireless Communications and Networking 1, 1–28 (2012).
3. M. Wolf and D. Kreß, “Short-range wireless infrared transmission: the link budget compared to RF,” IEEE Wireless Communications 10(2), 8–14 (2003). [CrossRef]
4. A.K. Majumdar and J.C. Ricklin, Free-space Laser Communications: Principles and Advances (Springer, 2008, vol. 2). [CrossRef]
5. J. Rzasa, S. Milner, and C.C. Davis, “Design and implementation of pan-tilt FSO transceiver gimbals for real-time compensation of platform disturbances using a secondary control network,” SPIE Optical Engineering+ Applications , 8162, 81620E (2011).
6. F.G. Walther, G. A. Nowak, S. Michael, R. Parenti, J. Roth, J. Taylor, W. Wilcox, R. Murphy, J. Greco, and J. Peters et al., “Air-to-ground lasercom system demonstration,” in Military Comm. Conference, (IEEE, 2010), pp. 1594–1600.
7. J. Rzasa, M. C. Ertem, and C. C. Davis., “Pointing, acquisition, and tracking considerations for mobile directional wireless communications systems,” SPIE Optical Engineering+ Applications, 88740C (2013).
8. T. Komine and M. Nakagawa, “Fundamental analysis for visible-light communication system using LED lights,” IEEE Transactions on Consumer Electronics 50(1), 100–107 (2004). [CrossRef]
9. J. Barry, Wireless Infrared Communications (Springer, 1994). [CrossRef]
10. Jelena Grubor, Sebastian Randel, Klaus-Dieter Langer, and Joachim W Walewski, “Broadband information broadcasting using LED-based interior lighting,” Journal of Lightwave Technology 26, 3883–3892 (2008). [CrossRef]
11. D. O’Brien, G. Parry, and P. Stavrinou, “Optical hotspots speed up wireless communication,” Nature Photonics 1, 245–247 (2007). [CrossRef]
12. H. Chowdhury and M. Katz, “Data download on move in indoor hybrid (radio-optical) WLAN-VLC hotspot coverages,” in Vehicular Technology Conference, (IEEE, 2013), 1–5.
13. T. Ho, S. Trisno, A. Desai, J. Llorca, S. Milner, and C. C. Davis, “Performance and analysis of reconfigurable hybrid FSO/RF wireless networks,” Proc. SPIE 119, 119–130 (2005). [CrossRef]
14. B. Epple, “Using a GPS-aided inertial system for coarse-pointing of free-space optical communication terminals,” Proc. SPIE 6304, 630418 (2006). [CrossRef]
15. L. Stotts, P. Cherry, P. Klodzy, R. Phillips, and D. Young, “Hybrid optical RF airborne communications,” in Proceedings of the IEEE (IEEE, 2009), pp. 1109–1127. [CrossRef]
16. S. Milner and C.C. Davis, “Hybrid free space optical/RF networks for tactical operations,” in Military Communications Conference, (IEEE, 2004), pp. 409–415.
17. H. Izadpanah, T. Elbatt, V. Kukshya, F. Dolezal, and B. K. Ryu, “High-availability free space optical and RF hybrid wireless networks,” IEEE Wireless Communications 10, 45–53 (2003). [CrossRef]
18. Y. Tang and M. Brandt-Pearce, “Link allocation, routing and scheduling of FSO augmented RF wireless mesh networks,” in International Conference on Communications, (IEEE, 2012), pp. 3139–3143.
19. I. S. Ansari, M.S. Alouini, and F. Yilmaz, “On the performance of hybrid RF and RF/FSO fixed gain dual-hop transmission systems,” in Electronics, Communications and Photonics Conference, (SIECPC, 2013), pp. 1–6.
20. W. Zhang, S. Hranilovic, and C. Shi, “Soft-switching hybrid FSO/RF links using short-length raptor codes: design and implementation,” IEEE Journal on Selected Areas in Communications 27, 1698–1708 (2009). [CrossRef]
21. M. Shur and R. Zukauskas, “Solid-state lighting: toward superior illumination,” in Proceedings of the IEEE, (IEEE, 2005), pp. 1691–1703. [CrossRef]
22. S. Pimputkar, J. Speck, S. DenBaars, and S. Nakamura, “Prospects for LED lighting,” Nature Photonics 3, 180–182 (2009). [CrossRef]
23. J. Kahn and J. Barry, “Wireless infrared communications,” in Proceedings of the IEEE, (IEEE, 1997), pp. 265–298. [CrossRef]
25. C. Chow, C. Yeh, Y. Liu, and P. Huang, “Mitigation of optical background noise in light-emitting diode (LED) optical wireless communication systems,” IEEE Photonics Journal 5, 7900307 (2013). [CrossRef]
26. A.J. Moreira, R.T Valadas, and A. de Oliveira Duarte, “Optical interference produced by artificial light,” Wireless Networks 3, 131–140, 1997. [CrossRef]
27. J. Kahn, J. Barry, M. Audeh, J. Carruthers, W. Krause, and G. Marsh, “Non-directed infrared links for high-capacity wireless LANs,” IEEE Personal Communications1, 12–25 (1994). [CrossRef]
28. D. OBrien, M. Katz, P. Wang, K. Kalliojarvi, S. Arnon, M. Matsumoto, R.J. Green, and S. Jivkova, “Short-range optical wireless communications,” Wireless World Research Forum, 1–22 (2005).
29. J. Grubor, S.C.J. Lee, K. Langer, T. Koonen, and J. Walewski, “Wireless high-speed data transmission with phosphorescent white-light LEDs,” in European Conference and Exhibition of Optical Communication-Post-Deadline Papers, (ECOC, 2008), pp. 1–2.
30. D. O’Brien and M. Katz, “Optical wireless communications within fourth-generation wireless systems,” Journal of Optical Networking 4, 312–322 (2005). [CrossRef]
31. M. Bilgi and Y. Murat, “Multi-element free-space-optical spherical structures with intermittent connectivity patterns,” in INFOCOM Workshops, (IEEE, 2008), pp. 1–4.
32. M. Yuksel, J. Akella, S. Kalyanaraman, and P. Dutta, “Free-space-optical mobile ad hoc networks: Auto-configurable building blocks,” Wireless Networks 15(3), 295–312 (2009). [CrossRef]