Inter-cell interference mitigation in multi-cellular visible light communications

Sun-Young Jung; Do-Hoon Kwon; Se-Hoon Yang; Sang-Kook Han

doi:10.1364/OE.24.008512

1. Introduction

Data traffic requirements in wireless communications have been explosively increased, and this tendency will be going on due to various applications for 5th-Generation wireless access (5G) like Internet-of-Things (IoT) and Machine Type Communications (MTC) [1, 2]. To support expanding traffic volume within a limited frequency resources, a spectral efficient transmission has been used in wireless access. Orthogonal Frequency Division Multiplexing (OFDM) is a highly spectral efficient parallel transmission technique based on orthogonally overlapped subcarriers, which is a standard of Long Term Evolution-Advanced (LTE-A) in wireless communications [3]. In addition, cell size in wireless access is getting smaller to support a huge traffic volume, which is so-called small cell, pico-cell, or femto-cell [4].

Light Emitting Diode (LED)-based Visible Light Communication (VLC) has been considered as an attractive research topic. Because there is no interference between VLC and existing wireless transmission, capacity could be effectively overlaid without wireless frequency resources. It could simultaneously provide lighting and communications with energy efficient eco technology. Thus, VLC could be utilized in indoor wireless transmission to supplement capability which is so-called Light Fidelity (Li-Fi) [5]. Moreover, VLC-based indoor positioning system is expected as an effective candidate with high accuracy for Localization-Based Service (LBS) because Global Positioning System (GPS) does not work inside buildings. In LED-based VLC, capacity improvement is one of the main issues to support an explosively increased traffic requirement in wireless services. Because common lighting LED has a small 3-dB bandwidth, many researchers have developed a high performance LED [6, 7] while employing spectrum efficient signaling, such as OFDM [8–12] in VLC with or without bandwidth extension techniques [13–15].

Primarily, research on VLC have been focused on a single channel transmission with identical broadcasting signal in every cell [8–12]. However, each cell should modulate different information to support various VLC applications, such as multiple heterogeneous services integrated Li-Fi transmission, VLC-based IoT, LBS and indoor positioning system. Furthermore, cell size will be getting smaller in Li-Fi transmission just like trend of wireless access. Recently, it has been reported in indoor VLC that average user net rate could be improved with a smaller cell size if handover rate is negligible [16]. In this situation, if adjacent lighting cells use same frequency band, inter-cell interference could be generated as illustrated in Fig. 1, which hinders receiver to detect proper signal and leads to performance degradation. In order to avoid the inter-cell interference, cell plan was proposed by employing RF carrier allocation [17] and frequency band allocation [18]. Although these techniques could mitigate inter-cell interference, a quite large frequency guard was required between adjacent cells, which reduces total capacity and Spectral Efficiency (SE). It may hinder LED-VLC from supporting a high data traffic, or it may require lighting LED to have higher performance which increases system cost. Moreover, these techniques require additional RF components for frequency up-conversion, which will make system to be complex and generate additional effect due to Carrier Frequency Offset (CFO).

Fig. 1 Inter-cell interference in VLC based multi-cellular Li-Fi system. Subset illustrates multicarrier-based cell partitioning distinguished by assigning subcarriers to each cell.

Download Full Size | PDF

In order to ensure high SE while using commercial lighting LED which has a small 3-dB bandwidth, multicarrier-based cell partitioning, such as OFDM-based coordination, could be used for cell plan by allocating subcarriers to cells combined with frequency reuse. In this system, all subcarriers are assigned to different cells belonging to single OFDM band [OFDM-based Multiple Access (OFDMA)], that is not belong to multiple bands up-converted at different center frequency. For RF wireless communications, OFDMA-based cell planning is wasteful because it requires several times larger signal bandwidth compared to RF up-conversion of multiple OFDM bands. Thus, it is better for RF wireless system to transmit multiple OFDM bands by using RF up-conversion, although additional effect like CFO would be generated. However, OFDMA-based cell planning is very good for optical wireless communications because additional electro-optic conversion process is equally required with similar bandwidth either RF up-conversion is included or not. Thus, OFDMA-based cell partitioning could make system to be simpler by eliminating up-conversion process and effect of CFO, while it could reduce frequency guard band between different cells in optical wireless multi-cellular transmission. Moreover, it could effectively support trilateration-based positioning method transparently [19] by using three subcarriers as pilots without frequency guard between the pilot and data subcarriers.

Even though the OFDMA-based cell partitioning could improve SE due to guard reduction, it still requires no small frequency guard between adjacent cells unless all cells are synchronized because asynchronous condition would change zero-crossing point of OFDM subcarriers and break orthogonality between cells. In other words, although a demanded guard band could be reduced compared to previous works [17, 18], several subcarriers are still required as a guard without tight synchronization even in OFDMA-based cell partitioning. Synchronization is a difficult task even in a single server because each cell has different electrical signal transmission path from central server, which requires timing advanced in transmitter [20] or Successive Interference Cancellation (SIC) in receiver [21, 22]. The timing advanced mandatorily requires large overhead for synchronization, and SIC is not perfect although having considerably large complexity due to several times repetition loop. Furthermore, synchronization problem will be more severe in multi-server system.

Filter Bank-based MultiCarrier (FBMC) is one of the multicarrier-based asynchronous waveforms. Recently, some asynchronous waveforms have been actively researched in wireless communications for future wireless access network [1, 2], but it is still new in optical communications excluding our previous works that experimentally evaluated in laser-based optical fiber transmission for passive optical network [23, 24]. Up to date, FBMC based on Mirabbasi-Martin equation shows the best sidelobes suppression among various known filters for multicarrier system within a given number of filter taps [25, 26].

In this paper, we propose multicarrier-based cell partitioning by using OFDMA in order to accommodate asynchronous multi-cellular Li-Fi transmission with high spectral efficiency. Moreover, it is experimentally demonstrated that inter-cell interference which was generated in OFDMA-based cell partitioning between asynchronous cells could be effectively mitigated by employing FBMC in proposed system. We experimentally verified that FBMC could effectively mitigate inter-cell interference as well as it improves SE and capacity in multicarrier-based cell partitioning for multi-cellular optical wireless communications. By using proposed multicarrier-based cell partitioning combined with FBMC, users could utilize interfered information transmitted from the other cells at inter-section, which could effectively increase total capacity and selection opportunity as well as support various applications.

2. Schematics

Figure 2 illustrates OFDM subcarrier allocation-based cell partitioning for asynchronous multi-cellular Li-Fi transmission in LED-VLC. A server generates multicarrier signals to support several different cells, or multiple servers could serve different cells. In order to achieve the proposed OFDMA-based cell partitioning, subcarriers are assigned to different cells to distinguish their own signal bandwidth as shown in Fig. 2. Bandwidth allocation could be flexible by changing the assigned subcarriers depending on data traffic demand. Essentially, subcarriers of OFDM could be presented as sinc-function due to rectangular time window. Subcarriers of OFDM are illustrated in Fig. 2(a) and 2(b), which is presented only a single subcarrier from each cell for simplicity. If received signals from different cells are perfectly synchronized, subcarriers of OFDM could be overlapped meanwhile satisfying orthogonal condition as illustrated in Fig. 2(a), which leads subcarriers to be independent to each other. However, when the server covers lighting units having unequal electrical signal path due to their location, which leads user to receive asynchronous signals transmitted from different cells. So, as illustrated in Fig. 2(b), if received signals are not synchronized, in a view point of Cell2 sample window, signals from Cell1 and Cell3 are received with insufficient symbol duration. A shortened symbol duration gives rise to spectrum stretching of corresponding subcarriers in frequency domain. Therefore, unless complex synchronization method [20–22] is employed, orthogonality would be broken and inter-cell interference would be generated. In order to mitigate inter-cell interference while employing OFDMA-based cell partitioning in asynchronous multi-cellular optical wireless communications, we used two method; Cyclic Prefix (CP) extension and FBMC.

Fig. 2 Schematics of OFDMA-based cell partitioning in multi-cellular asynchronous Li-Fi transmission based on LED-VLC.

Download Full Size | PDF

A. CP extension-based inter-cell interference reduction

Essentially, multipath fading robustness property of OFDM is originated from CP insertion. If delay spread is smaller than CP length, signals could be unaffected because whole OFDM symbol could be received due to CP. On the other hand, if delay spread is larger than CP length, a delayed signals are received with insufficient symbol duration in time domain, which makes subcarriers to be stretched within the signal bandwidth in frequency domain. Subcarriers of OFDM have sinc-shaped sidelobes, and its zero-crossing points are decided by signal bandwidth and number of subcarriers. The stretched subcarriers due to insufficient symbol duration change zero-crossing points, which leads spectrum leakage generating interference. Thus, in a commonly used OFDM, an inserted CP length is longer than delay spread to avoid performance degradation caused by multi-path fading channel. In multi-cellular Li-Fi system, electrical path difference may cause timing offset between asynchronous cells although we exclude optical delay. Therefore, CP extension with a proper length in OFDMA would reduce effect of inter-cell interference.

However, in OFDMA-based cell partitioning, timing offset effect is different from multipath fading of OFDM signal. Figure 3 shows asynchronously received signals at receiver when CP extension is employed in OFDMA-based cell partitioning.

Fig. 3 Asynchronously received signals with CP extension in OFDMA-based cell partitioning. An illustration presents simplified spectrum: (a) in Cell1 window, (b) in Cell2 window.

Download Full Size | PDF

x_{n}^{i} = \frac{1}{N} \sum_{k \in S i} X_{k}^{i} e^{\frac{j 2 π n k}{N}}, - N_{g} \leq n \leq N - 1

In Eq. (1), $X_{k}^{1}$ is frequency domain signal for kth subcarrier of ith cell. After N point IFFT process, $x_{n}^{i}$ is nth time domain sample of ith cell. $N_{g}$ is CP duration and Si is the assigned subcarrier index for ith cell. For simplicity, we just consider i = {1, 2}. In Fig. 3, $μ$ is timing offset between asynchronous cells. To reduce inter-cell interference between cells, CP extension is employed with an enough length ( $0 < μ \leq N_{g}$ ).

To focus interference caused by timing offset, transmission channel includes only timing offset without time/frequency gain/loss and additive noise. Thus, received signal after N point FFT process could be expressed,

\begin{array}{l} Y_{k}^{1} = N \sum_{n = 0}^{N - 1} \begin{matrix} (\frac{1}{N} \sum_{k \in S 1} X_{k}^{1} e^{\frac{j 2 π n k}{N}} + \frac{1}{N} \sum_{k \in S 2} X_{k}^{2} e^{\frac{j 2 π (n - μ) k}{N}}) e^{\frac{- j 2 π n k}{N}} \end{matrix} \\ = X_{k}^{1} + \sum_{n = 0}^{N - 1} \begin{matrix} (\sum_{k \in S 2} X_{k}^{2} e^{\frac{j 2 π n k}{N}} e^{\frac{- j 2 π n k}{N}} e^{\frac{- j 2 π μ k}{N}}) \end{matrix} \end{array}

where

Y_{k}^{1}

is received signal at kth subcarrier in Cell1 sample window. In Eq. (2), although delay is appeared, the signal transmitted from Cell2 could maintain its symbol duration due to CP. Therefore,

\sum_{n = 0}^{N - 1} \begin{matrix} e^{j 2 π n k / N} e^{- j 2 π n k / N} \end{matrix}

satisfies orthogonal condition, and

\sum_{k \in S 2}^{} X_{k}^{2} e^{- j 2 π n k / N} = 0

if

k \in S 1

(due to subcarrier allocation,

S 1 \cap S 2 = \emptyset

). Therefore, the received signal in Cell1 sample window is

Y_{k}^{1} = X_{k}^{1}

, which means that Cell1 signal could be demodulated without any interference. Thus, a proper length of CP could reduce inter-cell interference in Cell1 side. With the same manner, the received signals in Cell2 sample window could be expressed as

\begin{array}{l} Y_{k}^{2} = X_{k}^{2} + \frac{N - μ}{N} \sum_{n = 0}^{N - μ - 1} (\sum_{k \in S 1} X_{k}^{1} e^{\frac{j 2 π k n}{N}} e^{\frac{- j 2 π n k}{N}} e^{\frac{j 2 π k μ}{N}}) \\ + \frac{μ}{N} \sum_{n = N - μ}^{N - 1} (\sum_{k \in S 1} X_{k}^{1^{'}} e^{\frac{j 2 π k n}{N}} e^{\frac{- j 2 π n k}{N}} e^{\frac{- j 2 π k (N - μ)}{N}}) \end{array}

In right-hand side of Eq. (3), the second term and third term have different OFDM symbol due to timing offset;

X_{k}^{1^{'}}

is next symbol of

X_{k}^{1}

. In Eq. (3),

\sum_{n = 0}^{N - μ - 1} \sum_{k \in S 1} e^{j 2 π k n / N} e^{- j 2 π n k / N}

and

\sum_{n = N - μ}^{N - 1} \sum_{k \in S 1} e^{j 2 π k n / N} e^{- j 2 π n k / N}

could not maintain orthogonal condition due to insufficient cycle. Therefore, in right-hand side of Eq. (3), the second/third term remain as interference in Cell2 signal. Thus, although CP-extended OFDM was employed to reduce inter-cell interference, the performance improvement would be limited.

B. FBMC-based inter-cell interference mitigation

As mentioned above, interference of OFDM would be originated from zero-crossing points of sidelobes, which breaks orthogonal condition between overlapped subcarriers and generates spectrum leakage. To maintain orthogonality, a tight synchronization is required, which excessively increases system burden in terms of feedback overhead and complexity. Instead of a tight synchronization, it is possible to eliminate origin of interference by suppressing sidelobes of OFDM. To suppress sidelobes and mitigate inter-cell interference in multicarrier-based cell partitioning for multi-cellular optical wireless communications, a FBMC is employed [23, 24].

Figure 4 describes modulation/demodulation process of FBMC by comparing to those of OFDM. A basic structure of FBMC was the same to OFDM excepting CP insertion and digital filtering. To realize digital filter, frequency domain oversampling is required, which increases FFT size by several times in a number of filter taps. In order to maintain the same FFT size compared to OFDM, PolyPhase Network (PPN) [25] was employed in FBMC, which replaces frequency domain oversampling to time domain repetition. Maintaining the same FFT size will lead FBMC to have similar complexity with that of OFDM because FFT process occupies the highest computational complexity in signal generation.

Fig. 4 Modulation/demodulation process of OFDM and FBMC. Illustrations show characteristics of generated signals in frequency/time domain.

Download Full Size | PDF

Filtering increases symbol duration by several times depending on filter length decided by overlap factor (K). Moreover, although FBMC-based filtering could effectively suppress sidelobes, it breaks orthogonality between adjacent subcarriers. In order to avoid these filtering effect, Offset QAM (OQAM) with a half-symbol-duration offset and an overlap & sum process were employed in FBMC to ensure orthogonality between adjacent subcarriers while maintaining total capacity compared to OFDM. Thus, computational complexity and capacity of FBMC could be comparable to those of OFDM. As illustrated in Fig. 4, both spectrum sidelobes and pulse edge power will be effectively suppressed by using FBMC. Therefore, while mitigating inter-cell interference by eliminating a major cause, FBMC could be robust against asynchronous transmission even without CP, which could more improve spectral efficiency.

3. Experiments

In LED-based asynchronous multi-cellular optical wireless communications, experiments are conducted in two phases. Firstly, we analyze interference by assuming two adjacent cells without LED equalizing circuit at transmitter. Secondly, we add an equalizing circuit to extend 3-dB bandwidth of LED while assuming three adjacent cells for general applications.

A. Two cells without LED equalizer

Figure 5 presents experimental setup for accommodating multiple services in LED-based Li-Fi system. Arbitrary Waveform Generator (AWG) acts as a central server generating OFDMA signal, and each port of AWG produces different waveform for different cells. Output of AWG are electrically amplified/attenuated to optimize RF power. Additional RF cable is inserted in LED2 side to make asynchronous situation. Without considering of Wavelength Division Multiplexing (WDM), two commercial white LEDs (OSTAR LUW W5AM) are used to serve as asynchronous different cells. Bias current of LED is 100 mA. Convex lens is utilized to focus light because of a low optical power, but it could be eliminated by using arrayed LEDs. Because the LED has a spectral peak at ~440 nm, blue filter is inserted in front of APD to eliminate yellow phosphor component of white LED. Received signals are demodulated based on an offline process after sampling at Digital Phosphor Oscilloscope (DPO). The used signal bandwidth is 50 MHz, which was approximately ten times larger than LED 3-dB bandwidth. In our work, it is not possible to vary the receiver location due to a small optical power because we used single LED with convex lens. Thus, RF attenuators are used in order to emulate variation of receiver location as shown in inset of Fig. 5. The value (−5, −5) means relative signal power (dB) of LEDs according to the receiver location, which is controlled by RF attenuators. Number of subcarriers is 128 and first two subcarriers are nulled to avoid DC component. The LED2 uses 34th-95th subcarriers, and LED1 uses remaining subcarriers (3rd-32nd, 97th-128th subcarriers) excluding frequency guard (33rd, 96th). This type of allocation is used to obtain a similar average capacity because the channel response of LED is quite different according to frequency. Total assigned number of subcarriers is the same in two cells as 62.

Fig. 5 Experimental setup to emulate inter-cell interference between two different cells according to the receiver location.

Download Full Size | PDF

Frequency coefficients of filter in FBMC are 1, 0.971960, , and 0.235147 when an overlapping factor (K) is 4. These are converted to time coefficients by using IFFT to realize PPN. For intensity modulation, Hermitian symmetry is used in both OFDM and FBMC. For CP-extended OFDM, the inserted CP length is 4 samples (1/64 of symbol duration), whereas FBMC has no CP.

In order to equalize transmission channel effect, a simple single tap equalizer is used in both OFDM and FBMC. To realize adaptive equalization process, 4-preambles are inserted every 100-symbol to estimate channel state information. The channel Error Vector Magnitude (EVM) is calculated based on preambles and it is adaptively updated based on decision feedback until next preambles are received. Moreover, we employ an adaptive modulation to maximize total capacity in both OFDM and FBMC. Preambles are also used in this process. Based on feedback channel EVM information, bit-loading is performed at transmitter. In this work, bit-loading profile is not updated once it is decided because channel response is dominated by a quite stable LED frequency response.

B. Three cells with LED equalizer

In order to extend 3-dB bandwidth of LED, equalizing circuits are inserted before electro-optic conversion as shown in Fig. 6. The equalizer is consisted of adjustable capacitor and resistor, which controls low frequency response by playing as a High Pass Filter (HPF). It is worth to analyze the effect on multi-cellular system caused by equalizer because many researches have used this type of LED equalizer to overcome the limitation of a small bandwidth of LED. Basic structure for experiment is equal to a previous section (A) excepting the equalizer. To generalize cell planning scenario, three cells are considered because a cell partitioning based on a frequency reuse requires minimum three different frequencies to distinguish cells. In our work, three cells are emulated by two LEDs due to limitation of signal generator as illustrated in inset of Fig. 6. The LED1 plays as a role of Cell1 and Cell3, and LED2 serves Cell2 signal. By focusing on Cell2 signal, as receiver grows apart from Cell2, signal power would be decreased but power of Cell1 and Cell3 would have a smaller change compared to Cell2 signal power variation. Inversely, as receiver goes to center of Cell2, signal power would be increased but the power of Cell1 and Cell3 would have a smaller change because of adjacent other clusters. In other word, change of Cell1 and Cell3 signal power would not be equal to change of Cell2 signal power according to the receiver location. Thus, we evaluate performance by varying LED2 signal power when LED1 signal power is fixed. Moreover, this emulation is reasonable to focus on the effect of inter-cell interference by excluding the effect caused from degradation of Signal to Noise Ratio (SNR). RF attenuators are used to emulate receiver location as illustrated in inset of Fig. 6. Relative signal power of Cell1 and Cell3 are fixed as −5 dBm when relative Cell2 signal power is varied from −10 to 0 dBm by controlling RF attenuators. The used signal bandwidth is 150 MHz because the equalizer could extend LED 3-dB bandwidth. Bias current of LED is 50 mA to avoid an optical power saturation. Like the previous case, a number of subcarriers is 128 and first two subcarriers are nulled. The Cell1 and Cell3 are assigned by 3rd-40th and 87th-128th subcarriers, respectively. The Cell2 uses 42nd-85th subcarriers, and the remained subcarriers (41st, 86th) are assigned as guard interval.

Fig. 6 Experimental setup with equalizing circuit to extend 3-dB bandwidth of LED. Inset illustrates evaluation method to emulate interference according to the receiver location.

Download Full Size | PDF

Signal generation and filtering are similar to previous section (A), except for CP length of OFDM. The CP length is 8 samples (1/32 of symbol duration) in CP-extended OFDM, and FBMC still has no CP. The length of CP in OFDM is longer compared to preceding experiment, which is selected by considering a larger path difference for practical scenario. Channel estimation, equalization and adaptive modulation process are employed with the same manner to the preceding experiment. Table 1 summarizes the used parameters for signal generation.

Table 1. PARAMETER PROPERTIES FOR EXPERIMENTS

View Table

4. Results and discussion

A. Two cells without LED equalizer

Figures 7(a) and 7(b) show the measured RF spectra of Cell1 and Cell2 before adaptive modulation. In OFDM, sidelobes are obviously appeared above noise floor interfering each other as shown in Fig. 7(a). On the other hand, sidelobes could be effectively suppressed below noise floor by using FBMC as shown in Fig. 7(b). Because the used LED has very small 3-dB bandwidth, the response of integrated signals is rapidly decreased as frequency goes to higher in both cases. Primarily, because subcarrier of OFDM could be presented as sinc-function, a power of sidelobes would be exponentially decayed as it goes away from its main lobe, which leads most of inter-cell interference to be confined around boundary subcarriers as reported in our previous works [23, 24]. However, unlike to our previous works, sidelobes of Cell2 signal were amplified at low frequency in LED-VLC because low frequency region has a higher response compared to middle and high frequency region due to a small 3-dB bandwidth of LED. If a single cell or a perfectly synchronized multiple cells are used, the amplified sidelobes at low frequency will not affect the other subcarriers, but it is an impractical scenario in LED-based multi-cellular transmission. In other word, inter-cell interference will be generated not only at boundary subcarriers but also at low frequency subcarriers in OFDM due to the amplified sidelobes of the other cell, which would be even more severe in LED-VLC compared to our previous works. On the other hand, by employing FBMC, the amplified sidelobes are effectively suppressed and only small nonlinear components are generated in an optical modulation process, which are trivial to interfere the other cells although the same LED is used, as shown in Fig. 7(b).

Fig. 7 Integrated RF spectra of Cell1 and Cell2 before adaptive modulation: (a) in OFDM and (b) in FBMC.

Download Full Size | PDF

In Fig. 8, channel EVM of subcarriers are presented by emulating a receiver location. At center of inter-cell area (−5,-5), the received signal power from Cell1 and Cell2 are equal, so interference would be balanced as long as LEDs have a similar response. As receiver goes to center of Cell1 (0,-10), a received signal power from Cell1 is increased while Cell2 signal power is decreased. Signal to noise ratio of Cell2 signal is degraded due to a small RF power, and Signal to Interference Ratio (SIR) of Cell2 signal is also degraded due to increased Cell1 signal power. Thus, as illustrated in Fig. 8, channel EVM of Cell2 subcarriers goes worse as receiver moves from inter-section to Cell1. As shown in Figs. 8(b) and 8(c), Cell2 subcarriers experience SIR degradation as well as SNR degradation in OFDM. On the other hand, by comparing Figs. 8(a)-8(c), inter-cell interference is obviously mitigated, and Cell2 subcarriers experience only SNR degradation in FBMC.

Fig. 8 EVM of preamble by varying the receiver location: (a) FBMC, (b) OFDM without CP, and (c) CP-OFDM (4 samples)

Download Full Size | PDF

In case of OFDM, when receiver moves from inter-section to center of Cell2, the Cell1 subcarriers similarly experience SNR and SIR degradation, but there is different feature between Figs. 8(b) and 8(c). As shown in Fig. 8(c), EVM of Cell1 subcarriers are improved compared to Fig. 8(b) by inserting CP. In this work, the Cell1 signal early arrives at receiver, and Cell2 signal becomes a later one because the RF path difference is inserted in Cell2 side to make asynchronous signals. Without CP, as shown in Fig. 8(b), timing offset causes inter-cell interference at both sides due to stretched subcarriers generated by insufficient symbol duration. In CP-extended OFDM, if the timing offset is smaller than CP length, Cell2 signal is receives an enough symbol duration due to CP in Cell1 sample window, which enables asynchronous cells to maintain orthogonality. However, in Cell2 window, even with an enough CP, Cell1 signal arrives with CP of next symbol, which leads Cell1 subcarriers to be stretched in Cell2 window generating inter-cell interference as mentioned in Eq. (2)-(3). For this reason, only Cell1 could be improved even in CP-extended OFDM. Thus, inserting CP in OFDM could reduce inter-cell interference but it is not enough. In this work, one sample duration was 5 ns because sampling rate of AWG was 200 MHz, and propagation velocity in the used RF cable was 5 ns/m. Thus, it could be roughly estimated that RF path difference between cells is smaller than 4 m in this work, because interference could be reduced in Cell1 by inserting a 4-sample CP.

Figure 9 presents bit-loading profiles after adaptive modulation, which is decided based on a channel EVM presented in Fig. 8. As shown in Fig. 9(a), at the center of inter-cell, bit loading profile is continuous at boundary subcarriers in both OFDM and FBMC because cells have similar SNR and SIR. By using FBMC, subcarriers could carry more bits while ensuring the same signal quality (BER<~3.8 × 10⁻³) due to interference mitigation. From Fig. 9(b), we can obviously observe that only Cell1 subcarriers of OFDM (the early arrived one) could reduce interference by inserting CP. So, bit loading profile is not continuous in CP-extended OFDM, which is consistent to results in Fig. 8. Figures 10(a) and 10(b) present capacity of Cell1 and Cell2, respectively, according to receiver location. Capacity could be increased as the receiver goes close to a certain dedicated cell. In other word, Cell1 signal capacity is increased as the receiver moves from center of Cell2 (−10, 0) to center of Cell1 (0, −10), while Cell2 signal capacity is decreased. All of the cases in Fig. 10 have the same tendency; FBMC has the best performance in terms of capacity by avoiding inter-cell interference, and the worst case is OFDM without CP among three cases due to a large interference. Especially, in CP-inserted OFDM, there is difference between Figs. 10(a) and 10(b). Capacity of CP-OFDM is approaching to that of FBMC in Fig. 10(a), but there is no improvement in Fig. 10(b). Again, only an early arrived signal could be improved by inserting CP in OFDM, and a lately arrived signal has no improvement even with CP. Moreover, CP length should be changed according to system because required CP length depends on RF path difference. Thus, CP insertion is not a good choice to mitigate inter-cell interference in OFDMA-based cell partitioning for asynchronous multi-cellular Li-Fi transmission.

Fig. 9 Bit-loading profiles at center of inter-cell area: (a) FBMC and OFDM, (b) FBMC and CP-extended OFDM

Download Full Size | PDF

Fig. 10 Throughput variations according to the receiver location: (a) Cell1 and (b) Cell2.

Download Full Size | PDF

Figure 11 shows total capacity of Cell1 and Cell2 after adaptive modulation according to receiver location, where FBMC shows a better performance compared to OFDM. The throughput enhancement of FBMC is resulted from inter-cell interference mitigation. At center of inter-cell (−5, −5), total capacity of FBMC, OFDM, and CP-OFDM are 278, 191, and 227 Mbps, respectively. The capacity was evaluated by considering CP overhead in OFDM and filter overhead in FBMC. Because the used signal bandwidth was only 50 MHz, the corresponding SE are 5.7, 3.9, and 4.6 bit/s/Hz, respectively. The SE is evaluated by including guard subcarriers. Spectral efficiency is more effective measure than capacity because this work focuses on interference reduction with a relatively small signal bandwidth. By using FBMC, capacity and SE are improved by 1.5 times compared to those of OFDM. Moreover, when there is a large timing offset between asynchronous cells, SE improvement will be more critical issue factor because OFDM requires a longer CP.

Fig. 11 Total capacity of Cell1 and Cell2 according to the receiver location.

Download Full Size | PDF

B. Three cells with LED equalizer

In order to generalize cell planning and LED-based Li-Fi transmission, we set up three cells case with an equalizing circuit. In Fig. 12, frequency response of LED is presented with/ without equalizer. The used LED was commercial lighting element which has a small 3-dB bandwidth. As referred earlier, a commercial LED has especially large response at low frequency region, which amplifies sidelobes power and generates inter-cell interference at low frequency subcarriers far away from boundary subcarrier, unlike in our previous works. By adding equalizer, low frequency response is suppressed and the response of relatively higher frequency region becomes larger as shown in Fig. 12.

Fig. 12 Frequency responses of used LED with/without equalizer.

Download Full Size | PDF

Figures 13(a)-13(d) present spectra of OFDM and FBMC when three cells were integrated. At center of inter-cell (−5, −5), spectra of OFDM and FBMC are shown in Figs. 13(a) and 13(b), respectively. In Figs. 13(a) and 13(b), low frequency of Cell3 signal is increased due to Inter-Modulation Distortion (IMD). Electro-optic conversion process of LED has a nonlinear response. Moreover, by inserting equalizing circuit and reducing clipping ratio, the modulated signals experience more nonlinear distortion, which generates subcarrier-to-subcarrier beating in receiver due to direct detection. Lots of IMD components between subcarriers would be generated at around center frequency like Cell3 signal. In Fig. 13(a), low frequency of Cell2 signal becomes larger than that of Cell3, which comes from enhanced sidelobes of Cell2 signal summed up with IMD. These increased spectrum interferes to Cell1 signal with a broken orthogonality because cells asynchronously transmit their own signals. Moreover, sidelobes of Cell1 and Cell3 signal affect to Cell2 signal. On the other hand, in case of FBMC, low frequency distortion components of Cell2 signal are remained the same compared to Cell3 signal as presented in Fig. 13(b). In other word, a small IMD is observed but the amplified sidelobes of Cell2 is not observed in FBMC due to sidelobes suppression, which could mitigate inter-cell interference, not only in boundary subcarriers but also in low frequency region.

Fig. 13 Spectra of OFDM and FBMC when three cells were served. (a) OFDM at inter-cell. (b) FBMC at inter-cell. (c) OFDM at Cell2. (d) FBMC at Cell2.

Download Full Size | PDF

Additionally, spectra of OFDM and FBMC are presented in Figs. 13(c) and 13(d), respectively, when the receiver located at center of Cell2 (−5, 0). In both cases, Cell2 signal power are increased by 5 dB, and interference from Cell2 is increased. The increased signal power would promote generation of IMD components, in terms of amplitude and spectral width. In OFDM, as shown in Fig. 13(c), Cell2 signal spectrum is changed compared to Fig. 13(a) due to sum of amplified sidelobes and broadened IMD, which increases inter-cell interference to other cells. However, as shown in Fig. 13(d), IMD components are negligibly increased in FBMC although Cell2 power was increased because of sidelobes suppression.

Figures 14(a) and 14(b) present bit-loading profiles at center of inter-cell and center of Cell2, respectively, after adaptive modulation. The amount of inter-cell interference could be inferred from violation of spectrum in Fig. 13, which deteriorates EVM of preamble initially modulated by 4QAM. In our work, because the bit-loading is realized by using feedback information of channel EVM based on preamble, subcarrier which experienced interference could not load a large number of bits. Thus, bit-loading profiles in Fig. 14 well matches with spectra of Fig. 13. Normally, in a single cell or a perfectly synchronized multiple cells transmission scenario, spectrum overlap is not an issue because orthogonal condition will be maintained as mentioned earlier. Moreover, interference between asynchronous cells is commonly confined around their boundary subcarrier in OFDMA. Thus, in normal OFDMA, signal quality could be ensured by inserting frequency guard around boundary subcarriers although it reduces efficiency in terms of spectrum utilization. However, in LED-based asynchronous multi-cellular transmission, the effect of inter-cell interference becomes more serious because of a rapidly degraded frequency response and nonlinearity of lighting LED. The growth of sidelobes and IMD provide a performance degradation not only in boundary subcarriers but also in whole signal bandwidth, which would be more severe when receiver goes away from a dedicated cell, and so the bit-loading profile is changed as shown in Fig. 14.

Fig. 14 Bit-loading profiles after adaptive modulation: (a) at center of inter-cell, (b) at center of Cell2.

Download Full Size | PDF

Total capacity and SE are presented in Figs. 15(a) and 15(b), respectively, by varying Cell2 signal power when Cell1 and Cell3 signal powers are fixed to emulate generalized OFDMA-based cell partitioning with three asynchronous cells. As the receiver goes to center of Cell2, capacity of Cell2 signal is increased in both cases (FBMC_2 and OFDM_2) due to SNR improvement, which is shown in Fig. 15(a). By using FBMC-based interference mitigation, total capacity and SE could be improved by gap between OFDM and FBMC in Figs. 15(a) and 15(b). If a dedicated cell is Cell1 or Cell3, capacity will become smaller as the receiver goes to center of Cell2 (FBMC_1 + 3 and OFDM_1 + 3) due to SIR and SNR degradation. Signal to noise ratio degradation could be excluded because Cell1 and Cell3 signal power were fixed only to focus on the effect of interference. Capacity of Cell1 and Cell3 signals in OFDM are reduced as increasing Cell2 signal power due to amplified sidelobes and IMD, while only a trivial degradation is observed in FBMC due to a small IMD without amplified sidelobes. For various applications, like handover process, Device-to-Device (D2D), and Co-ordinated Multi-Point (CoMP), the receiver would require signals transmitted from adjacent cells. In this case, although SNR of certain cell increased in OFDMA-based cell partitioning, total capacity and SE would be saturated because inter-cell interference more deteriorates signal performance of adjacent cells as shown in Fig. 15(b). On the other hand, by using FBMC, inter-cell interference could be effectively mitigated and the following total capacity and SE could be improved as increasing SNR of certain cell.

Fig. 15 Total capacity and SE according to Cell2 signal powers when Cell1 and Cell3 signal power were fixed in three cells scenario. (a) Total capacity. (b) Spectral efficiency.

Download Full Size | PDF

5. Conclusion

We proposed multicarrier-based cell partitioning to accommodate multi-cellular optical wireless communications based on OFDMA. By assigning subcarriers to different LED lighting cells, one could reduce frequency guard band and make a simpler system by eliminating additional RF components. Moreover, it is experimentally demonstrated that inter-cell interference is generated between asynchronous lighting cells in OFDMA-based cell partitioning, and it is more severe in LED-based Li-Fi system because interference affect not only boundary subcarriers but also low frequency subcarriers due to amplified sidelobes and IMD. In order to reduce inter-cell interference generated in multicarrier-based cell partitioning, FBMC is employed, and performance improvement is experimentally verified by comparing capacity. By using FBMC, inter-cell interference between asynchronous lighting cells could be effectively mitigated, which leads to improvement in total capacity and spectral efficiency. Moreover, by using FBMC in multicarrier-based cell partitioning, it could effectively support trilateration-based indoor positioning system while providing an increased cell selection opportunity. Thus, we believe that multicarrier-based cell partitioning combined with FBMC could be one of the strong solutions to support asynchronous multi-cellular transmission for future optical wireless communications.

Acknowledgments

This work was supported by the ICT R&D programs of MSIP/IITP, Republic of Korea. [R0101-16-0086, Research on 25Gbps optical access network based on discrete multitoned enabling dynamic network resource management].

References and links

1. G. Wunder, P. Jung, M. Kasparick, T. Wild, F. Schaich, Y. Chen, S. Brink, I. Gaspar, N. Michailow, A. Festag, L. Mendes, N. Cassiau, D. Kténas, M. Dryjanski, S. Pietrzyk, B. Eged, P. Vago, and F. Wiedmann, “5GNOW: non-orthogonal, asynchronous waveforms for future mobile applications,” IEEE Commun. Mag. 52(2), 97–105 (2014). [CrossRef]

2. G. Wunder, M. Kasparick, S. Brink, F. Schaich, T. Wild, I. Gaspar, E. Ohlmer, S. Krone, N. Michailow, A. Navarro, G. Fettweis, D. Ktenas, V. Berg, M. Dryjanski, S. Pietrzyk, and B. Eged, “5GNOW: challenging the LTE design paradigms of orthogonality and synchronicity,” http://arxiv.org/abs/1212.4034

3. E. Dahlman, S. Parkvall, and J. Skold, 4G: LTE/LTE-Advanced for Mobile Broadband, 2nd ed. (Academic, 2015), Ch. 1.

4. T. Nakamura, S. Nagata, A. Benjebbour, Y. Kishiyama, T. Hai, S. Xiaodong, Y. Ning, and L. Nan, “Trends in small cell enhancements in LTE advanced,” IEEE Commun. Mag. 51(2), 98–105 (2013). [CrossRef]

5. D. Tsonev, S. Videv, and H. Haas, “Light fidelity (Li-Fi): towards all-optical networking,” Proc. SPIE 9007, 900702 (2014).

6. D. Tsonev, H. Chun, S. Rajbhandari, J. J. D. McKendry, S. Videv, E. Gu, M. Haji, S. Watson, A. E. Kelly, G. Faulkner, M. D. Dawson, H. Haas, and D. O’Brien, “A 3-Gb/s single-LED OFDM-based wireless VLC link using a gallium nitride μLED,” IEEE Photonics Technol. Lett. 26(7), 637–640 (2014). [CrossRef]

7. H. Chun, P. Manousiadis, S. Rajbhandari, E. A. Vithanage, G. Faulkner, D. Tsonev, J. J. D. McKendry, S. Videv, E. Xie, E. Gu, M. D. Dawson, H. Hass, G. A. Turnbull, I. D. W. Samuel, and D. C. O’Brien, “Visible light communication using a blue GaN μLED and fluorescent polymer color converter,” IEEE Photonics Technol. Lett. 26(20), 2035–2038 (2014). [CrossRef]

8. J. Y. Sung, C. W. Chow, and C. H. Yeh, “Dimming-discrete-multi-tone (DMT) for simultaneous color control and high speed visible light communication,” Opt. Express 22(7), 7538–7543 (2014). [CrossRef] [PubMed]

9. J. Vucic, C. Kottke, S. Nerreter, K. D. Langer, and J. W. Walewski, “513 Mbit/s visible light communications link based on DMT-modulation of a white LED,” J. Lightwave Technol. 28(24), 3512–3518 (2010).

10. G. Cossu, A. M. Khalid, P. Choudhury, R. Corsini, and E. Ciaramella, “3.4 Gbit/s visible optical wireless transmission based on RGB LED,” Opt. Express 20(26), B501–B506 (2012). [CrossRef] [PubMed]

11. T. Komine, S. Haruyama, and M. Nakagawa, “Performance evaluation of narrowband OFDM on integrated system of power line communication and visible light wireless communication,” in Proceedings of 1st International Symposium on Wireless Pervasive Computing (IEEE, 2006), pp.1–6. [CrossRef]

12. A. M. Khalid, G. Cossu, R. Corsini, R. Choudhury, and E. Ciaramella, “1-Gb/s transmission over a phosphorescent white LED by using rate-adaptive discrete multitone modulation,” IEEE Photonics J. 4(5), 1465–1473 (2012). [CrossRef]

13. H. L. Minh, D. O’Brien, G. Faulkner, L. Zeng, K. Lee, D. Jung, Y. Oh, and E. T. Wo, “100-Mb/s NRZ visible light communications using a postequalized white LED,” IEEE Photonics Technol. Lett. 21(15), 1063–1065 (2009). [CrossRef]

14. H. Li, X. Chen, B. Huang, D. Tang, and H. Chen, “High bandwidth visible light communications based on a post-equalization circuit,” IEEE Photonics Technol. Lett. 26(2), 119–122 (2014). [CrossRef]

15. H. L. Minh, D. O’Brien, G. Faulkner, L. Zeng, K. Lee, D. Jung, and Y. Oh, “High-speed visible light communications using multiple-resonant equalization,” IEEE Photonics Technol. Lett. 20(14), 1243–1245 (2008). [CrossRef]

16. S. Pergoloni, M. Biagi, S. Colonnese, R. Cusani, and G. Scarano, “Optimized LEDs footprinting for indoor visible light communication networks,” IEEE Photonics Technol. Lett. 28(4), 532–535 (2016). [CrossRef]

17. H. S. Kim, D. R. Kim, S. H. Yang, Y. H. Son, and S. K. Han, “Mitigation of inter-cell interference utilizing carrier allocation in visible light communication system,” IEEE Commun. Lett. 16(4), 526–529 (2012). [CrossRef]

18. D.-R. Kim, S.-H. Yang, S.-K. Han, Y.-H. Son, and H.-S. Kim, “Single side-band orthogonal frequency division multiplexing signal transmission in RF carrier allocated visible light communication,” IET Optoelectron. 7(6), 125–130 (2013). [CrossRef]

19. H. S. Kim, D. R. Kim, S. H. Yang, Y. H. Son, and S. K. Han, “An indoor visible light communication positioning system using a RF carrier allocation technique,” J. Lightwave Technol. 31(1), 134–144 (2013). [CrossRef]

20. Evolved Universal Terrestrial Radio Access (E-UTRA); Medium Access Control (MAC) protocol specification, 3GPP TS 36.321.

21. T. Cover, “Broadcast channels,” IEEE Trans. Inf. Theory 18(1), 2–14 (1972). [CrossRef]

22. D. Tse and P. Viswanath, Fundamentals of Wireless Communication (Cambridge University Press, 2005)

23. S. Y. Jung, S. M. Jung, H. J. Park, and S. K. Han, “Mitigation of timing offset effect in IM/DD based OFDMA-PON uplink multiple access,” Opt. Express 23(11), 13889–13898 (2015). [CrossRef] [PubMed]

24. S. Y. Jung, S. M. Jung, and S. K. Han, “AMO-FBMC for asynchronous heterogeneous signal integrated optical transmission,” IEEE Photonics Technol. Lett. 27(2), 133–136 (2015). [CrossRef]

25. M. Bellanger, D. LeRuyet, D. Roviras, M. Terré, J. Nossek, L. Baltar, Q. Bai, D. Waldhauser, M. Renfors, T. Ihalainen, A. Viholainen, T. H. Stitz, J. Louveaux, A. Ikhlef, V. Ringset, H. Rustad, M. Najar, C. Bader, M. Payaro, D. Katselis, E. Kofidis, L. Merakos, A. Merentitis, N. Passas, A. Rontogiannis, S. Theodoridis, D. Triantafyllopoulou, D. Tsolkas, D. Xenakis, M. Tanda, T. Fusco, M. Huchard, J. Vandermot, A. Kuzminskiy, F. Schaich, P. Leclair, and A. Zhao, “FBMC Physical Layer: A Primer,” http://www.ict-phydyas.org.

26. A. Sahin, I. Guvenc, and H. Arslan, “A survey on multicarrier communications: prototype filters, lattice structures, and implementation aspects,” IEEE Comm. Surv. and Tutor. 16(3), 1312–1338 (2014). [CrossRef]

Parameter	A: Two cells without LED equalizer	B: three cells with LED equalizer
Signal bandwidth (MHz)	50	150
Subcarrier spacing (KHz)	390.625	1171.875
Clipping ratio (dB)	10	5
CP length	OFDM: 0, 4 / FBMC: 0	OFDM: 8 / FBMC: 0
AWG sampling rate (MHz)	200	600
LED bias current (mA)	100	50
Subcarriers (effective)	128 (124)
IFFT/FFT size	512 (Hermitian symmetry and oversampling)
Initial modulation	4 QAM
Filter shape	OFDM: Rectangular / FBMC: Mirabbasi-Martin (K = 4)

Inter-cell interference mitigation in multi-cellular visible light communications

Abstract

1. Introduction

2. Schematics

A. CP extension-based inter-cell interference reduction

B. FBMC-based inter-cell interference mitigation

3. Experiments

A. Two cells without LED equalizer

B. Three cells with LED equalizer

4. Results and discussion

A. Two cells without LED equalizer

B. Three cells with LED equalizer

5. Conclusion

Acknowledgments

References and links

Cited By

Figures (15)

Tables (1)

Equations (3)

Optics Express