WO2025041258A1 - Occ transmitter, occ receiver, occ signal generation method, occ signal reception method, and program - Google Patents

Abstract

The present invention provides an OCC transmitter comprising a dimmer and an on–off keying modulation unit. The dimmer converts a high-rate signal on the basis of a low-rate signal by m-out-of-n encoding, and generates a superimposed signal that gives two desired dimming levels. Provided that Y₀ and Y₁ denote the two dimming levels, respectively, m and n are related by m = Y₀n and m = Y₁n. The on–off keying modulation unit converts the signal outputted by the dimmer into an optical signal. An OCC receiver comprises a low-speed camera, a RoI identification unit, a high-speed camera, an on–off keying demodulation unit, and an inverse dimmer. The low-speed camera captures an image of the OCC signal and outputs a low-rate signal. The RoI identification unit identifies a RoI from the low-rate signal. The high-speed camera extracts the RoI from the camera field of view and outputs a RoI image capture signal. The on–off keying demodulation unit demodulates the RoI image capture signal and acquires a superimposed signal. The inverse dimmer uses a codebook to decode the superimposed signal into a high-rate signal.

Description

OCC transmitter, OCC receiver, OCC signal generating method, OCC signal receiving method, and program

The disclosed technology relates to dimming control and encoding technology for optical camera communication.

Optical camera communication (OCC) is a visible light communication method that uses a light source, such as an LED display, as the transmitting means and a camera as the receiving means. The transmitting side transmits a digital signal by modulating the light emitted by a light source such as an LED. Modulation is performed by controlling the light source to be turned on and off.

The IEEE 802.15.7m group (TG 7m) introduced visible light Region-of-Interest (RoI) signaling into vehicular OCC systems (Non-Patent Document 1). This technology allows OCC systems to transmit low-rate and high-rate data streams simultaneously.

While the low-rate stream is used for RoI identification, the high-rate stream is used for high-speed data communication over the selected RoI. Hybrid modulation schemes such as Twinkle Variable Pulse Position Modulation (VPPM) and Hybrid Spatial Phase-Shift Keying (HS-PSK) are often used to modulate the low-rate and high-rate streams.

Specifically, the high-rate stream is modulated by Twinkle VPPM (Non-Patent Document 2) or Dimming Spatial-8 Phase-Shift Keying (DS8-PSK) (Non-Patent Document 1), and the low-rate stream is modulated by Undersampled Frequency-Shift On-Off Keying (UFSOOK), Undersampled Phase-Shift On-Off Keying (UPSOOK), or Spatial-2 Phase-Shift Keying (S2-PSK).

Both streams transmitted from the RoI are decoded by the dual-camera receiver, where the high-speed (HS) camera demodulates the high-rate stream and the low-speed (LS) camera demodulates the low-rate stream. This reduces the computational load on the high-speed camera-based receiver.

M. D. Thieu et al, “Optical RoI-signaling for vehicular communications”, IEEE Access, vol. 7, pp.69873 - 69891, 2019. R. Roberts, “Intel OCC Proposal”, IEEE Standards Assoc., Jan. 2016, [Retrieved July 25, 2020], Internet <https://bit.ly/3dUL7ZK> D.-P. Nguyen et al., "A probabilistic shaping approach for optical region-of-interest signaling", IEEE Photonics Technology Letters, vol. 34, no. 6, pp. 309 - 312, Mar. 2022. T. Fehenberger et al., “Multiset-partition distribution matching”, IEEE Transactions on Communications 67, no. 3 (2018), pp.1885-1893.

Dimming control is one of the key issues in OCC and visible light communication (VLC) systems.
Traditional approaches use correction symbol insertion and puncturing for dimming control, which adds a large number of redundant bits to the frame, reducing the achievable transmission rate.

The dimming level can be managed by adjusting the proportion of the optical bit stream that is OFF (hereafter referred to as "logic 0") or ON (hereafter referred to as "logic 1") between 0% and 100%. Dimming control in HS-PSK and Twinkle VPPM involves changing the duty cycle (length of ON or OFF) of the high-rate pulse groups. This makes demodulation difficult for DS8-PSK and Twinkle VPPM signals, which require each optical pulse to be synchronized, sampled, and decoded with strict timing.

Furthermore, in traditional VLC systems, long sequences of logical 1s or 0s are to be avoided because they cause flicker and clock data recovery (CDR) problems. For this reason, run-length-limited (RLL) codes maintain direct current (DC) balance by having an equal number of logical 1s and logical 0s in each symbol.

There is a trade-off between rate loss and ease of implementation in RLL codes such as Manchester, 4B6B, and 8B10B.
Furthermore, when it comes to forward error correction (FEC), systems that use both FEC and RLL only support hard-decision (HD) FEC decoders, as most RLL decoders generate hard output information, which results in poor error correction performance.

The soft-input soft-output (SISO) RLL decoder introduced by Kim et al. improves system reliability by generating soft outputs for the soft-decision (SD) FEC decoding algorithm. However, the decoding algorithm for SISO RLL codes is complex and cannot be integrated into the VLC receiver.

Instead of using hybrid modulation techniques with the drawbacks mentioned above, Nguyen et al. proposed probabilistic shaping for RoI signaling applications (Non-Patent Document 3), but implementation remains challenging.

In order to solve the above problem, an OCC transmitter according to the disclosed technology includes a dimmer and an on-off keying modulation unit.
The dimmer converts the high-rate signal based on the low-rate signal using m-out-of-n encoding to generate a superimposed signal that provides the two desired dimming levels.
The m and n in the m-out-of-n coding are expressed as m=Y0n, m=Y1n, where Y0 and Y1 are the two dimming levels, respectively.
A codebook for m-out-of-n encoding is generated using a set of bit sequences Cm having m logical 1s or logical 0s in n bits, and pairs Cm-l and Cm+l (l is a natural number).
The on-off keying modulator converts the signal output from the dimmer into an optical signal.
In addition, an OCC receiver according to the disclosed technology, which receives an OCC signal transmitted by an OCC transmitter, includes a low-speed camera, an RoI identification unit, a high-speed camera, an on-off keying demodulation unit, and an inverse dimmer.
The low-speed camera captures the OCC signal and outputs a low-rate signal.
The RoI determination unit determines the RoI from the low rate signal.
The high-speed camera extracts the RoI from the camera's field of view and outputs an RoI shooting signal.
The on-off keying demodulation unit demodulates the RoI imaging signal to obtain a superimposed signal.
The inverse dimmer uses a codebook to decode the high rate signal from the superimposed signal.

Thanks to the non-RLL feature, soft-decision forward error correction algorithms can be applied, which increases the reliability of the system.
The disclosed 4p-EMDPM dimmer-based visible light RoI signaling technique enables easier clock tracking and synchronization compared to the conventional twinkle VPPM approach.
The disclosed 4p-EMDPM method reduces rate loss (bit redundancy) due to probabilistic shaping more than conventional methods.

FIG. 2 is a functional block diagram of an OCC transmitter and an OCC receiver according to an embodiment. A comparison of rate loss for different distribution matching schemes. A comparison of rate losses among MPDM, EMPDM, 2p-EMPDM, and 4p-EMPDM. Functional block diagram of the 4p-EMPDM dimmer and codebook structure diagram. A diagram showing the codebook structures of MPDM, EMPDM, and 4p-EMPDM. A diagram showing the binary tree structure of 4p-EMPDM. 4 is a diagram for explaining details of the flow of digital signal processing in an OCC transmitter and an OCC receiver according to an embodiment. FIG. Photograph of the low-speed RoI signaling-high-speed OCC demonstration system showing the OCC transmitter, low-speed camera, and video monitor. A photo of the low-speed RoI signaling-high-speed OCC demonstration system showing the OCC transmitter, high-speed camera, and OCC receiver. The relationship between exposure time and pixel Eb/N ₀ was measured by changing the distance between the transmitter and receiver. Diagram showing how noise affects the intensity I and area A of a 50% dimming symbol. Diagram showing how noise affects the intensity I and area A of the 30% dimming symbol and the 70% dimming symbol. FIG. 1 shows two-class classification using intensity domain thresholding. A diagram showing the codebook sizes of the 4p-EMPDM dimmer and the EMPDM dimmer with Υ ₀ = 41% and Υ ₁ = 59%. Diagram showing 4p-EMPDM dimming and FEC encoding overhead. Figure showing the output probability distribution of 100,000 information words shaped by the 4p-EMPDM dimmer and encoded by the SPE according to the LPS architecture. Diagram showing the setup of the low-rate and high-rate experimental systems. FIG. 4 is a flowchart illustrating the operation of the OCC transmitter 1. FIG. 4 is a flowchart illustrating the operation of the OCC receiver 2. FIG. 2 is a diagram showing an example of the functional configuration of a computer.

Below, an embodiment of the disclosed technology will be described in detail. Components having the same functions will be given the same numbers, and duplicate explanations will be omitted.

The disclosed technology provides binary extended multiset-partition distribution matching (EMPDM) with four composition pairs (4p), and a dimmer based on 4p-EMPDM for visible light RoI signaling applications. We first explain the disclosed technology through an overview of the proposed system in which 4p-EMPDM is used, followed by a theoretical analysis of the proposed algorithm and an experimental optimization to find the optimal parameters of a 4p-EMPDM dimmer dedicated for RoI signaling applications.

[Outline of the proposed system]
The functional block diagram of the proposed system is shown in Figure 1. The two inputs to the system are a low-rate signal v and a high-rate signal μ = (μ ₁ , μ ₂ , ..., μ _k ), where v and μ _j (1≦j≦k) take the values 0 or 1.
Let the proportion of logic 1s in an n-bit 4p-EMPDM dimming codeword u and a z-bit FEC codeword y be type PA(1)(%) and ω(%), respectively. In the OCC transmitter, the 4p-EMPDM dimmer encodes an information word μ with a Bernoulli (1/2) distribution into a codeword u = (u ₁ , u ₂ , ..., u _n ) with an arbitrary distribution.

In fact, when the message u is encoded by a systematic FEC encoder, ω will have a slightly different value than PA(1). Following the Layered Probabilistic Shaping (LPS) architecture, we use a systematic polar encoder (SPE) to preserve the probabilistic shape of ω and generate two dimming levels in the non-RLL VLC transmitter. Furthermore, instead of using an RLL code to limit the run length as in traditional VLC systems, the 4p-EMPDM dimmer shapes the information word μ such that the proportion of logical 1s and logical 0s is not significantly different. As a result, the run length of y is limited to the range that does not cause flicker at the minimum optical clock rate.
In this specification, the frequency of occurrence of logic 1 and logic 0 in codeword u (e.g., "1011101011" has "three 0s and seven 1s") is called a "composition." Also, a composition with a frequency of occurrence of logic 1 i is represented as γ _i . For example, the sequence "1011101011" belongs to γ ₇ .

In the dual camera receiver (OCC receiver), a low-speed global shutter camera with a frame rate of 30 fps and long exposure time detects the dimming level of a Light Emission Diode (LED), which decodes the logic 0s and 1s in a low-rate bitstream that is used for RoI identification.
The high-speed camera decodes the high-rate data stream on the RoI detected by the low-speed camera at a high frame rate (e.g., 10,000 fps). The SD successive canceling SPD (SD-SPD) decodes the codeword y' into u'. Finally, the inverse 4p-EMPDM dimmer decodes the message u' to reconstruct the high-speed binary sequence μ', assuming that all transmission errors have been corrected by SD-SPD.

The modulation technique chosen was OOK due to its simplicity and popularity in the VLC system.
Since the code word u is a combination of a low-rate signal and a high-rate signal, the code word u and the code word u' before decoding (inverse dimming) are sometimes referred to as "superimposed signals" in this specification.

ここで

は、不等式

を満たす最大のｋを出力する。 [Binary Extended MPDM (EMPDM)]
The matching rate R _DM , which indicates how many bits of information are transmitted with each bit coded in binary DM, is defined by the following equation (1).

where

is an inequality

Then, the largest k that satisfies is output.

バイナリDMが達成しうる最大マッチング率は式（３）となる。

バイナリDMのレート損失は、式（４）のように定義される。

The entropy of a discrete random variable A of type P _A with alphabet B = {0,1} is given by equation (2).

The maximum matching rate that a binary DM can achieve is given by equation (3).

The rate loss for binary DM is defined as in equation (4).

The rate loss of Constant Composition Distribution Matching (CCDM) increases as the block length decreases (Reference 1).

Reference 1: P. Schulte et al., "Constant composition distribution matching", IEEE Trans. Inf. Theory, vol. 62, no. 1, pp.430 - 434, Jan. 2016.

MPDM was introduced as a method to reduce rate loss, especially for short block lengths, and is suitable for practical implementation (Non-Patent Document 4). The rate loss reduction of MPDM comes from the inclusion of multiple Composition Pairs (CPs) in addition to a Typical Composition (TC) to generate a non-fixed composition DM. The compositions used in a pair are complementary to each other, such that the probabilistic average of the sequences generated by both compositions achieves the desired distribution.

Non-Patent Document 4 introduced a binary tree structure in which each information word is divided into a payload that is mapped to a binary CCDM sequence and a prefix that selects the composition.
Unlike conventional MPDM, which requires the sequence number of a particular composition to be rounded to the nearest power of 2 (power of 2 constraint), the disclosed technology proposes binary EMPDM, which lifts the power of 2 constraint for each composition and extends the codebook of conventional matching. In addition, to make EMPDM implementable with a binary tree structure, we propose sharing the codebook between the TC and the top CP, and between adjacent compositions.

　TCとすべてのCPを含むEMPDMコードブックの順列の総数は、次のように与えられる。

ここで、N_{ｐａｉｒｓ}は、制約{γ_ｍ－ｌ, γ_ｍ＋ｌ}=2{γ_ｍ}(lは自然数)を満たす有効なCPの数を示し、m<n/2のとき最大でm、m>n/2のとき最大でn-mである。 Cm represents a binary group of n elements of weight m, forming the codebook of TC. For low-level dimming, m= _Υ0 ·n, and for high-level dimming, m= _Υ1 ·n. Let |C _i | denote the number of elements in C _i , then the number of permutations in the codebook of composition pairs {γ _m−l , γ _m+l } (l>0) can be calculated as follows:

The total number of permutations in the EMPDM codebook including TC and all CPs is given by:

Here, N _pairs denotes the number of valid CPs that satisfy the constraint {γ _m-l , γ _m+l }=2{γ _m } (l is a natural number), and is at most m when m<n/2, and at most nm when m>n/2.

　同様に、Υ_１>0.5の場合、min(|C_ｍ－ｌ|,|C_ｍ＋ｌ|)=|C_ｍ＋ｌ|であり、式（５）（６）より式（８）を得る。

　低レベル用シェーピングΥ_０と、高レベル用シェーピングΥ_１のEMPDMコードブックのシーケンス数は次のとおりである。

When Υ ₀ <0.5, min(|C _m−l |, |C _m+l |)=|C _m−l |, and from equations (5) and (6),

Similarly, when Υ ₁ >0.5, min(|C _m−l |, |C _m+l |)=|C _m+l |, and equation (8) is obtained from equations (5) and (6).

The number of sequences in the EMPDM codebook for low-level shaping Υ ₀ and high-level shaping Υ ₁ is as follows:

EMPDM is a fixed-length invertible mapping, so the number of input bits k and the output sequence length n are fixed. Imposing the constraint Υ ₀ = 1 - Υ ₁ leads to an implementation that supports two symmetric shaping levels for Rol signaling applications. Imposing this constraint on equation (9) gives

From (1) and (9), the matching rate of EMPDM at two shaping levels can be written as follows:

The rate loss of EMPDM, Δ _EMPDM , is estimated by replacing R _EMPDM in equation (10) with R _DM in equation (4). Figure 2 shows the rate loss of different distribution matching schemes. The sequences are shaped for different power distributions PA(1) from 1% to 99%.
It can be seen that the rate loss of EMPDM is much lower than that of other shaping methods for all power distributions. Moreover, under the constraint Υ ₀ = 1 - Υ ₁ , the rate loss is symmetric with respect to PA(1) = 50%. The symmetry of the matching rate and rate loss is important in designing a DM-based dimmer that allows two dimming levels with the same rate loss.

Now, it appears that implementing EMPDM using look up tables (LUTs) would be difficult if the EMPDM codebook contained all CPs. Furthermore, since flicker is a fundamental problem in non-RLL VLC systems, shaped sequences with long free runs of logical 1s or 0s need to be avoided. Therefore, the remainder of the disclosed technique optimizes EMPDM to make it compatible with VLC-based RoI signaling applications.

[4p-EMPDM dimmer and its implementation method]
For RoI signaling applications, we propose a 4p-EMPDM dimmer that supports two dimming levels _{, Υ0} and _Υ1 , where _Υ0 and _Υ1 are the typical power distributions of the dimmer for logic 0 and logic 1 of the low-rate stream, respectively.
Assuming that 1≦m<n, to see how many times larger C _m is than C _m-1 , calculate the following formula.

<Example> When n=100, _C25 is 452.5 times larger than _C20 and 957293 times larger than _C15 . In other words, _C15 << _C20 << _C25 .

According to equation (11), most of the nonuniform sequences in the EMPDM codebook have TC(γ _m ) and dominant CP{γ _m−l , γ _m+l }, where l is a small positive value.
Figure 3 shows the rate loss for several examples of MPDM and EMPDM, including codebooks with TC and two leading CPs (2p-EMPDM), four leading CPs (4p-EMPDM), and all CPs (EMPDM).
4p-EMPDM is a compact version of EMPDM that generates only run-length aware codewords (codewords with a maximum run-length bounded to a predefined value) for the purpose of reducing flicker in VLC applications. Arrow 31 shows the empirical shaping range for low-level dimming (logic 0) and arrow 32 shows the empirical shaping range for high-level dimming (logic 1).

The gap around PA(1)=50% is provided to ensure that detection of the logic 0 and logic 1 states of low-rate signals is not unclear. The run-length-aware shaping range for VLC applications is marked with a shaded box. Comparing EMPDM in Figure 2 with 4p-EMPDM in Figure 3, we can see that the rate loss of 4p-EMPDM is equivalent to that of EMPDM. In particular, there is no difference in the range of PA(1)<45% at low level dimming and in the range of PA(1)>55% at high level dimming.

On the other hand, the rate loss of 2p-EMPDM increases as a penalty for not including the third and fourth CPs in the codebook. In a non-RLL VLC system, the codeword needs to be shaped so that there is no significant difference between the occurrence ratios of logic 1 and logic 0. If the two dimming levels are around 50% (e.g., _Υ0 = 41%, _Υ1 = 59%), there should be little difference between the occurrence ratios of logic 1 and logic 0. Therefore, the lowest rate loss and DC balance are the two criteria for determining the constrained shaping range of the 4p-EMPDM dimmer. After experimental evaluation, two run-length-aware shaping ranges were determined: 37%<PA(1)<45% and 55%<PA(1)<63% (see Figure 3).

Figure 4 shows a detailed functional block diagram and codebook structure of the 4p-EMPDM dimmer 11 in Figure 1. The 4p-EMPDM dimmer supports two dimming levels with the same rate loss through symmetry with respect to type PA(1)=50%, as described in [Binary Extended MPDM (EMPDM)].
The low-level 4p-EMPDM dimming unit 41 and the high-level 4p-EMPDM dimming unit 42 respectively use a codebook of composition pairs (C _m-l , C _m+l ) (where l={1, 2, ₃ , 4}) and a codebook of typical compositions (C _m ) according to m determined from the dimming levels Υ ₀ and Υ 1, and the rate loss is smaller than that of CCDM.

Figure 5 shows the codebook structures of conventional MPDM, EMPDM, and 4p-EMPDM. 4p-EMPDM is a compact version of EMPDM with a typical composition γ _m and four composition pairs {γ _m-l , γ _m+l } (l={1, 2, 3, 4}). As explained in [Binary Extended MPDM (EMPDM)], EMPDM is an extension of the MPDM codebook because its composition codebook does not have a power-of-two constraint. The 4p-EMPDM codebook has a binary tree structure, and adjacent compositions share the codebook. In addition, the 4p-EMPDM codebook contains only four CPs in addition to the TC. This is because the size of the codebook for the remaining CPs is not significant compared to the size of the codebook for the four CPs.

ただし、

は、端数処理して最も近い２の累乗とする関数を表す。 From (7) and (8), the total number of output sequences of the two-level 4p-EMPDM dimmer is

however,

represents a function that rounds to the nearest power of two.

Figure 6 shows the binary tree structure of 4p-EMPDM. A Huffman code is used to determine the prefix that identifies the subset pair. The first p _l bits (1 or 2 bits) are used for the prefix. The next bit (the first bit of the payload) selects the shared subset in the subset pair. In this specification, the prefix and the first bit of the payload are sometimes collectively referred to as the "selection code."

There are two types of shared subsets.
(i) Shared exemplar-pairwise subset: The codebook of a TC is shared with the codebook of its paired composition element, e.g., the subset with (C _m , C _m−1 ) and (C _m , C _m+1 ).
(ii) Shared Pairwise Subsets: A codebook is shared between two pairs of composition elements, e.g., the subsets with (Cm _-1 , Cm _-2 ), (Cm ₊₁ , Cm ₊₂ ), (Cm _-2 , Cm _-3 , Cm _-4 ), and (Cm ₊₂ , Cm ₊₃ , Cm ₊₄ ). Each subset of leaf nodes applies binary CCDM based on m-out-of-n codes to map the remaining _kl- bit payload one-to-one to a shaped sequence.

[OCC transmitter]
Here, the OCC transmitter of the disclosed technology and its operation will be described with reference to FIG. 1 and FIG.
The OCC transmitter 1 includes a dimmer 11 (4p-EMPDM dimmer), an FEC encoding unit 12 (systematic polar encoding unit), an OOK modulation unit 13, and an LED 14.
Two dimming levels Υ ₀ and Υ ₁ are defined in advance, and m = Υ ₀ n at dimming level Υ ₀ and m = Υ ₁ n at dimming level Υ _1. A codebook for m-out-of-n encoding is created from C _m and pairs of C _m-l and C _m+l (l = 1, 2, 3, 4).
The low rate signal v and the high rate signal μ are input to the dimmer 11 (steps S1801 and S1802).
Based on the low-rate signal and the high-rate signal, the dimmer 11 converts the high-rate signal into an m-out-of-n code using the codebook, and outputs a superimposed signal (step S1803).
The FEC encoder 12 encodes the superimposed signal using, for example, a systematic polar code, and outputs an FEC code (step S1804).
The OOK modulator 13 modulates the FEC code to drive the LED 14 and output an OCC signal (step S1805).

[OCC receiver]
Next, the OCC receiver of the disclosed technology and its operation will be described with reference to FIG. 1 and FIG.
The OCC receiver 2 includes a low-speed camera 21, a high-speed camera 22, an RoI identifying unit 23, an OOK demodulating unit 24, an FEC decoding unit 25 (systematic polar decoding unit), and an inverse dimmer 26 (4p-EMPDM inverse dimmer).
The low-speed camera 21 captures the entire camera field of view and outputs it to the RoI identifying unit 23 (step S1901).
The RoI identifying unit 23 detects a low rate signal from the camera's field of view (step S1902).
The RoI identifying unit 23 identifies the RoI in the field of view from the detected low-rate signal (step S1903), and outputs the RoI information to the high-speed camera 22.
Based on the RoI information, the high-speed camera 22 extracts the RoI portion from the imaging signal (camera field of view) and outputs the RoI imaging signal (step S1904).
The OOK demodulation unit 24 demodulates the RoI imaging signal and outputs the FEC code (step S1905).
The FEC decoding unit 25 decodes the superimposed signal from the FEC code (step S1906).
The inverse dimmer 26 decodes the high-rate signal from the superimposed signal using the m-out-of-n encoding codebook generated in the same manner as the dimmer 11 (step S1907).

６:TCと４つのCPを、コードブックのシーケンス数に応じて降順にソートする。
７:シーケンスの総数を、シーケンス総数を超えない最も近い2のべき乗（2^ｋ）の値に丸める。ここで、kは調光器の入力ビット数である。
８:シーケンス総数が2^ｋ個のコードブックを、大きさを変えて分割してサブセットのペアに割り当てる。例えば、シェアド・典型－ペアワイズ・サブセットに2のk_ｉ乗個を、シェアド・ペアワイズ・サブセットに2のk_ｉ+1乗個を割り当てる（図５、図６を参照）。
９:サブセットのペアを識別するため、p_ｌビットをプレフィックスビットとして予約する。プレフィックスビットの次のビットは、ペア内のサブセットを識別するために使用される。
（ｉ）シェアド・典型－ペアワイズ・サブセットの場合:p_ｌ=1
（ｉｉ）シェアド・ペアワイズ・サブセットの場合:p_ｌ=2
10:m-out-of-nコードを使用して、k_ｌビットペイロードを、各サブセット内のシェイプされたシーケンスにマップする。 The details of the codebook construction of the 4p-EMPDM dimmer and the construction of its binary tree structure are as follows.
1: Find the output distributions Υ ₀ , Υ ₁ that produce the two desired dimming levels.
2: Calculate the typical composition γ _m for each dimming level. For low-level dimming, m = Υ ₀ · n, and for high-level dimming, m = Υ ₁ · n. The low-rate signal v determines the dimming level.
3: Find all CPs that satisfy the constraint {γ _m-1 , γ _m+1 }=2{γ _m } (where l>0).
4: Delete the unselected sequences in the larger of the composition pairs (C _m-l , C _m+l ). After this deletion, the number of sequences is |(C _m-l , C _m+l )|=2・min(|C _m-l |, |C _m+l |), where l={1, 2, 3, 4}.
5: The main codebook is restricted to TC and four CPs. The total number of sequences in the codebook is given by

6: Sort the TC and the four CPs in descending order according to the number of sequences in the codebook.
7: Round the total number of sequences to the nearest power of two (2 ^k ) that does not exceed the total number of sequences, where k is the number of input bits of the dimmer.
8: A codebook with a total of ^2k sequences is divided into subset pairs of varying sizes. For example, 2k _i sequences are assigned to the shared typical-pairwise subset, and 2k _i+1 sequences are assigned to the shared pairwise subset (see Figures 5 and 6).
9: To identify a pair of subsets, p _l bits are reserved as prefix bits. The bit next to the prefix bit is used to identify the subset within the pair.
(i) Shared representativeness - pairwise subset case: p _l =1
(ii) For shared pairwise subsets: p _l =2
10: Use an m-out-of-n code to map the k _l -bit payload to a shaped sequence in each subset.

[Experimental optimization]
In this section, we present a method to optimize the shaping level of the 4p-EMPDM dimmer through experimental evaluation. The system supports two dimming levels, representing logic 0 and logic 1 levels for low-rate communication in Rol signaling applications. Each time the dimmer changes the power distribution PA(1), the dimming level of the LEDs reflects the change. A low-speed camera-based receiver decodes the low-rate stream to identify the RoI by detecting the dimming level.

In experimental optimization, there are three criteria for finding the optimal shaping level of the dimmer:
(1) The rate loss is the lowest.
(2) The output distribution is within a range that takes into account the run length of non-RLL VLC (see Figure 3).
(3) The corresponding dimming levels can be easily detected and demodulated by a low-speed camera-based receiver.

FIG. 7 shows the flow of digital signal processing (DSP) for the proposed transmitter and receiver, and the experimental setup.
8 and 9 show the low-rate RoI signaling-high-rate OCC demonstration system.
FIG. 17 summarizes the setup of the low-rate and high-rate experimental systems.
The low-rate RoI signal receiver used a 30 fps low-speed global-shutter camera to detect the dimming level and decode the low-rate bitstream. An Atmos Shogun 7 with SSD mini-storage served as the video capture and monitoring device. VLC transmitters with low-power LEDs were placed at distances of 1.7 m, 2.9 m, and 4.2 m from the receiver.

On the other hand, we use a high-speed camera receiver (HS camera/OCC Receiver) that operates at 10,000 fps with an exposure time of 50 μs. Images captured by the high-speed camera are sent to a field programmable gate array (FPGA) for processing. In addition, to evaluate the performance of the proposed method, we built a simulation system of a high-speed OCC system on the OCC channel.

ここで、N_０はノイズ密度、E_ｂはビットエネルギー、θ＝T_ｅ/Tはカメラの露光時間とビット間隔の比率、aはマークとスペースの振幅、αとβフィッティングパラメータである。
　非特許文献１で実験的に推定したα=0.01529とβ=0.1973を用いて、pixel Eb/N_０を計算した。また、露光時間の長いカメラは、高周波信号を減衰させるローパスフィルターと考えることができる。そのため、露光時間を調整するとpixel Eb/N_０に大きく影響する。 How noise affects image quality has been widely studied in the literature. In general, the impact of noise varies from one imaging system to another. In OCC systems, noise is modeled as a Gaussian distribution. The main noise sources are thermal noise generated by the readout circuitry and shot noise due to background illumination. The variance of the shot noise is proportional to the area of the received pixel and the received light intensity. The pixel noise of a charge-coupled device/complementary metal-oxide semiconductor (CCD/CMOS) camera can be approximated as follows, assuming one bit per symbol:

where N ₀ is the noise density, E _b is the bit energy, θ = T _e /T is the ratio of the camera exposure time to the bit interval, a is the mark and space amplitude, and α and β are fitting parameters.
Pixel Eb/N ₀ was calculated using α = 0.01529 and β = 0.1973, which were experimentally estimated in Non-Patent Document 1. In addition, a camera with a long exposure time can be considered as a low-pass filter that attenuates high-frequency signals. Therefore, adjusting the exposure time has a large effect on pixel Eb/N ₀ .

As shown in Figure 10, pixel Eb/ _N0 changes with the exposure time of the CMOS camera. However, when _Te exceeds 4500μs, pixel Eb/ _N0 saturates at about 39.7dB at distances of 1.7m, 2.9m, and 4.2m. With a Tx clock rate of 20kHz, the average luminance of a 128-bit frame requires 6.4ms to be fully sensed by the low-speed camera. Therefore, to find the optimal shaping level of the 4p-EMPDM dimmer, we set _Te = 6400μs. Then, pixel Eb/ _N0 is 39.74dB.

Figure 11 shows how noise affects the intensity I and area A of a 50% dimming symbol. We obtain an intensity variance |maxI - minI| of 14.5 and an area variance |maxA - minA| of 10. We take the variance at the 50% dimming level as the within-class variance to determine the threshold range for demodulating the low-rate bitstream.

Conventional OCC systems determine the LED on/off state by comparing the intensity to a threshold. To determine the threshold, the VLC transmitter transmits 100,000 symbols shaped by a 4p-EMPDM dimmer with 30%<PA(1)<70%.
12 shows the received symbols shaped by the dimmer with Υ ₀ =30% for logic 0 and Υ ₁ =70% for logic 1. Only the intensity information is used to determine the logic 0/logic 1 decision threshold.
The exposure time of the camera sensor is _Te = 6400 μs. The camera-based receiver is placed 4.2 m away from the VLC transmitter. It can be seen that the pixel intensities of the symbols shaped with a dimmer of _Υ1 = 70% do not change significantly. The receiver can only detect the difference in pixel area for the symbols shaped with different compositions.

On the other hand, the symbols shaped with a dimmer of Υ ₀ = 30% show a much larger variation in intensity, because the pixel intensity reaches saturation at dimming levels close to 70%, and VPPM has been found to perform best at this dimming level (Non-Patent Document 1).
This finding revealed that in addition to pixel intensity, the pixel area of the received signal is useful for data demodulation. Therefore, our low-speed camera-based receiver uses the intensity and area information of the captured signal for data demodulation and threshold determination.

ここで、sとbはそれぞれしきい値線の勾配とy切片を表し、次の式（１５）、（１６）を使用して計算できる。

ここで、AとIは、それぞれキャプチャされたピクセルの領域（幅×高さ）と平均強度を表す。 Specifically, a low threshold, a median threshold, and a high threshold are used to determine a threshold range, which is the range between a low threshold and a high threshold. The within-class variance due to noise effects determines the width of the range. The threshold range is defined to avoid ambiguity in detecting close dimming levels. The median threshold is given by the following equation (14):

where s and b represent the slope and y-intercept of the threshold line, respectively, and can be calculated using the following equations (15) and (16).

where A and I represent the area (width × height) and average intensity of the captured pixels, respectively.

ただし、b_ｈとb_ｌは、式（１４）の中心しきい値の上下にある最も遠い点の座標を置き換えることによって決定される。 The high and low threshold lines can be determined using a linear equation.

where b _h and b _l are determined by substituting the coordinates of the furthest points above and below the central threshold in equation (14).

To check whether any point (X _i , Y _i ) is above or below the central threshold, insert the coordinates (X _i , Y _i ) of that point into the threshold line equation (14), and compare Y i with s.X+b.
If _Yi < s.X+b then the point is below the line, else if Yi > s.X+b then the point is above the threshold.

Two-class classification by intensity domain threshold is shown in Figure 13. The camera captures the dimming symbol shaped by the 4p-EMPDM dimmer. The typical shaping percentages are Υ ₀ = 41% for logic 0 and Υ ₁ = 59% for logic 1.
Both the intensity and area of a captured pixel are used to determine an intensity-area threshold. The exposure time of the camera sensor is T _e =6400 μs. The camera-based receiver is placed 4.2 m away from the VLC transmitter. Using the intensity-area threshold, a typical shaping level is close to 50%; specifically, 30% to 41% is a logical 0 and 70% to 59% is a logical 1.
As a result, the rate loss of the dimmer decreases from 0.031 to 0.0165 (Fig. 3), which means that the matching rate increases with the proposed area-intensity threshold.

Figure 14 shows the codebook sizes of the 4p-EMPDM and EMPDM dimmers with Υ ₀ = 41% and Υ ₁ = 59%. These optimal power distributions were experimentally determined using intensity-area thresholds to determine the low-rate logic levels. As most of the codebook belongs to the TC and four CPs, the codebook size of the 4p-EMPDM dimmer is found to be 90.88% of that of the EMPDM dimmer. In the ranges of PA(1) = [37%, 45%] and PA(1) = [55%, 63%], the rate loss of the 4p-EMPDM dimmer is comparable to that of the EMPDM dimmer (Figure 3). As explained in the section [4p-EMPDM Dimmer and Its Implementation Method], the 4p-EMPDM dimmer generates only sequences with no significant difference in the ratio of logic 0s and logic 1s. Therefore, compared to other binary distribution matching techniques, 4p-EMPDM is suitable for non-RLL VLC systems because the run length of its output sequence is constrained to mitigate flicker.

Figure 15 shows the overhead of the 4p-EMPDM dimmer and FEC coding. The coding overhead is the ratio of the number of redundant bits to the number of information bits and can be expressed as a percentage. The overhead of 4p-EMPDM is 4.16%, and the overhead of SPE(FEC) is 28%.

Although the systematic FEC encoder better preserves the output probability distribution of the 4p-EMPDM dimmer compared with the non-systematic FEC encoder, the logical operations on the generator and parity matrix of the systematic FEC encoder usually change the output distribution of the dimmer codewords. Therefore, we investigated the output distribution of the output sequence of the LPS in a Monte Carlo simulation. Figure 16 shows the output probability distribution of 100,000 information words shaped by the 4p-EMPDM dimmer and encoded by the SPE according to the LPS architecture. It can be seen that the SPE slightly brings the output distribution of the dimmer closer to PA(1)=50%.
For Rol signaling applications, a low-rate logic level is created from two dimming levels, therefore, the power distributions of the LPS symbols corresponding to the two dimming levels must not overlap in order to be identified by a low-speed camera-based receiver.

[Program, recording medium]
The various processes described above can be implemented by loading a program that executes each step of the above method into the recording unit 2020 of the computer 2000 shown in FIG. 20, and operating the control unit 2010, input unit 2030, output unit 2040, display unit 2050, etc.

The program describing this processing can be recorded on a computer-readable recording medium. Examples of computer-readable recording media include magnetic recording devices, optical disks, magneto-optical recording media, and semiconductor memories.

The program may be distributed, for example, by selling, transferring, or lending portable recording media such as DVDs or CD-ROMs on which the program is recorded. Furthermore, the program may be stored in a storage device of a server computer, and the program may be distributed by transferring the program from the server computer to other computers via a network.

A computer that executes such a program, for example, first stores in its own storage device the program recorded on a portable recording medium or the program transferred from a server computer. Then, when executing a process, the computer reads the program stored on its own recording medium and executes the process according to the read program. As another execution form of the program, the computer may read the program directly from the portable recording medium and execute the process according to the program, or may execute the process according to the received program each time a program is transferred from the server computer to the computer. The above-mentioned process may also be executed by a so-called ASP (Application Service Provider) type service that does not transfer the program from the server computer to the computer, but realizes the processing function only by issuing an execution instruction and obtaining the results. Note that the program in this form includes information used for processing by an electronic computer that is equivalent to a program (such as data that is not a direct command to the computer but has properties that specify the processing of the computer).

In addition, in this embodiment, the device is configured by executing a specific program on a computer, but at least a portion of the processing may be realized by hardware.

1 OCC transmitter 11 Dimmer 12 FEC encoder 13 OOK modulator 14 LED
15 Memory unit 2 OCC receiver 21 Low speed camera 22 High speed camera 23 RoI identification unit 24 OOK demodulation unit 25 FEC decoding unit 26 Inverse dimmer 27 Memory unit 2000 Computer 2010 Control unit 2020 Recording unit 2030 Input unit 2040 Output unit 2050 Display unit

Claims (9)

An OCC transmitter that generates an OCC signal in which a low rate signal and a high rate signal are superimposed,
a dimmer that converts a high-rate signal based on a low-rate signal by m-out-of-n encoding to generate a superimposed signal that provides two desired dimming levels;
an on-off keying modulation unit that converts the signal output by the dimmer into an optical signal;
the dimmer converts the high-rate signal by m-out-of-n coding to _output an n-bit signal, with the two dimming levels being Υ ₀ and Υ ₁ , _respectively , and with m = Υ ₀ n at the dimming level Υ 0 and m = Υ ₁ n at the dimming level Υ 1;
An OCC transmitter in which Cm is a set of bit sequences having m logical 1s or logical 0s in n bits, and the m-out-of-n encoding codebook is generated using Cm and pairs of Cm-l and Cm+l (l is a natural number up to m when m<n/2, and up to nm when m>n/2). An OCC transmitter that generates an OCC signal in which a low rate signal and a high rate signal are superimposed,
a dimmer that converts a high-rate signal based on a low-rate signal by m-out-of-n encoding to generate a superimposed signal that provides two desired dimming levels;
an on-off keying modulation unit that converts the signal output by the dimmer into an optical signal;
the dimmer converts the high-rate signal by m-out-of-n coding to _output an n-bit signal, with the two dimming levels being Υ ₀ and Υ ₁ , _respectively , and with m = Υ ₀ n at the dimming level Υ 0 and m = Υ ₁ n at the dimming level Υ 1;
An OCC transmitter in which Cm is a set of bit sequences having m logical 1s or logical 0s in n bits, and the codebook for the m-out-of-n encoding is generated using Cm and pairs of Cm-l and Cm+l (l = 1, 2, 3, 4). 3. The OCC transmitter of claim 2,
The codebooks are created as follows: one codebook CB0 is created from Cm, Cm-1, and Cm+1; one codebook CB100 is created from Cm-1 and Cm-2; one codebook CB101 is created from Cm+1 and Cm+2; one codebook CB110 is created from Cm-2, Cm-3, and Cm-4; and one codebook CB111 is created from Cm+2, Cm+3, and Cm+4.
OCC transmitter. 4. The OCC transmitter of claim 3,
the high rate signal has a selection code for selecting the codebook;
The dimmer selects the codebook CB0 when the selection code is 0, selects the codebook CB100 when the selection code is 100, selects the codebook CB101 when the selection code is 101, selects the codebook CB110 when the selection code is 110, and selects the codebook CB111 when the selection code is 111.
OCC transmitter. 2. An OCC receiver that receives an OCC signal generated by the OCC transmitter according to claim 1 and decodes the OCC signal into a low-rate signal and a high-rate signal,
A slow-speed camera
An RoI identification unit;
High-speed cameras and
An OOK demodulator;
an inverse dimmer;
The RoI determination unit obtains the low rate signal from the low speed camera and determines an RoI;
The high-speed camera extracts the RoI from the camera field of view and outputs an RoI imaging signal;
The OOK demodulation unit acquires the superimposed signal from the RoI imaging signal,
The inverse dimmer decodes the high rate signal from the superimposed signal using the codebook. 1. A method for generating an OCC signal having two dimming levels in which a low rate signal and a high rate signal are superimposed, comprising:
A step of generating an m-out-of-n encoding codebook by the dimmer using C _m and pairs of _{C m-l} and C _m+l (l is a natural number up to m when m<n/2, and up to nm when m>n/2), where C m is a set of bit sequences having m logical 1s among n bits;
The dimmer converts the k-bit signal into an m-out-of-n coding format, where the two dimming levels _are Υ ₀ and Υ ₁ , and m=Υ _{0 n for the dimming level Υ 0 and m=Υ 1 n} for the dimming level Υ 1, to generate a superposition signal;
An on-off keying modulation unit converts a signal output from the dimmer into an optical signal;
The OCC signal generating method includes: 7. An OCC signal receiving method for receiving an OCC signal generated by the OCC signal generating method according to claim 6 and decoding a high-rate signal, comprising the steps of:
A low-speed camera captures the OCC signal and outputs a low-rate signal;
A step of an RoI determination unit determining an RoI from the low rate signal;
A step in which a high-speed camera extracts the RoI from a camera field of view and outputs an RoI imaging signal;
an OOK demodulation unit demodulating the RoI imaging signal to obtain the superimposed signal;
an inverse dimmer decoding the high rate signal from the superimposed signal using the codebook;
2. A method for receiving an OCC signal, comprising: A program for causing a computer to function as an OCC transmitter according to any one of claims 1 to 4. A program for causing a computer to function as the OCC receiver described in claim 5.

Images