JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 21, NO.

11, NOVEMBER 2003

2557

Research of Algebra Congruent Codes Used in Two-Dimensional OCDMA System

Tao Pu, Student Member, IEEE, Yu Quan Li, and Shu Wen Yang
AbstractAfter summarizing the application of linear, quadratic, and hyperbolic operators in code-division multiple-access (CDMA) system, a generalized algebraic congruent code is presented by its construction formula in three different categories: frequency-hop (FH), time-spread (TS), and FH-TS codes. On the discussion of those three codes auto- and cross-correlations and their contribution to bit-error-rate (BER) performances, a novel two-dimensional optical CDMA (OCDMA) system is provided in which different types of codes are used to provide different types of service. Such an OCDMA system is very fit for the case that a small part of services are high bit rate or high quality of service, while a great amount of services are not. Index TermsAlgebraic congruent code (ACC), optical codedivision multiple-access (OCDMA).

I. INTRODUCTION PTICAL code-division multiple-access (OCDMA) technology could be used to provide random and asynchronous multi-access in optical networks. The full use of the all-optical process made it a high-speed, as well as protocol-transparent, method for multi-accessing and multiplexing. As for optical time-division multiple access (OTDMA), OCDMA is destined to be an essential supplement of wavelength-division multliple access (WDMA) in access networks as well as backbone networks. A one-dimensional (1-D) temporal OCDMA system always has limited codes capacity, spared sequences, long code length, and bad correlation. A two-dimensional (2-D) OCDMA system could overcome those shortcomings by using 2-D codes that have double orthogonality in temporal and frequency domain. Since arrayed-waveguide grating (AWG) [1], [2] and fiber grating [3], [4] could be used in OCDMA system to accomplish 2-D coding and decoding, more and more attention has been paid to this realm. 2-D address codes discussed by those papers were one-coincidence sequences [3], extended hyperbolic congruence (EHC) [4] code, prime sequence/optical orthogonal code (Ps/OOC) [5], etc. In fact, all the FH codes used in wireless FH communication systems could be used in the 2-D OCDMA system directly.
Manuscript received November 27, 2002; revised July 8, 2003. This work was supported by the National Science Foundation (NSF) of China. T. Pu and Y. Q. Li are with the Institute of Communication Engineering, Nanjing, Jiangsu 210016, China (e-mail: [email protected]; [email protected]). S. W. Yang is with Advanced Tech Research Center, ShenZhen University, Guangdong 518060, China (e-mail: [email protected]). Digital Object Identifier 10.1109/JLT.2003.819549

The algebraic congruent coding (ACC) method in wireless FH communication could be used in an OCDMA system to construct 2-D FH codes, 1-D TS sequences, and even 2-D FH-TS codes by combining codes in two domains. ACC means using all kinds of congruent operators within the Galois field to acquire address sequences that are needed in OCDMA system. With different operators, different sequences could be given, such as prime, quadratic congruent, hyperbolic congruent, and Costas sequences. Using TS, FH, and FH-TS codes from the same congruent operator as the address sequences of different users, services of variable data rates as well as variable quality of service (QoS) could be provided in a hybrid 2-D OCDMA system. Such a random asynchronous multi-rate multi-QoS and an all-optical process schematic should very beneficial to future networks. The paper is organized as followed. In Section II, the generation of three types of ACC (FH, TS, and FH-TS codes) are introduced to show their lengths, weights, self-properties, and capacities. The cross-properties between those ACC are demonstrated in Section III. The bit-error-rate (BER) performances due to multi-access interferences (MAIs) are analyzed in Section IV, and the conclusion could be drawn that the quasi-orthogonality of all users is kept, and a usable BER performance could be given in such a three-code hybrid system. II. THE CONSTRUCTION OF THREE CATEGORIES OF ACC AND THEIR PROPERTIES A. The Construction of ACC is used to represent a cerWith a given prime number , , tain algebraic congruent operator within Gelois field , is the index of sequences in a family, while points out the position of time slots in a se, we could define three types quence. With a given operator of ACC, i.e., FH, TS, and FH-TS codes. The three codes are expressed as follows: FH code:

(1) TS code:

(2)

0733-8724/03$17.00 2003 IEEE

2558

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 21, NO. 11, NOVEMBER 2003

FH-TS code:

TABLE I FH-TS CODES GENERATED BY p = 3 LINEAR CONGRUENT OPERATOR

(3) where represents modulo- multiplication. Each array in formulas (1)(3) represents the 1 codes time-wavelength position. From formulas (1)(3), we could observe that the former two ACCs capacities are both , while the FH-TS codes capacity . An example of FH-TS code construction is shown is is in Table I, where the linear congruent operator , there are codes used. Since the prime number in total. B. The Relationship Between ACCs We use Fig. 1 to illustrate the relationship between FH, TS, , we have and FH-TS codes. From a given operator and . TS sethe FH array shown in Fig. 1, where quence is generated by tearing apart the rows of the FH array and placing them in sequence. In FH-TS code array, the marks time positions are determined by TS sequence, while their wavelength position are determined by FH arrays marks line by line. For high QoS service, identical TS codes could be used upon wavelengths simultaneously; therefore, we called it parallel TS code. C. The Properties of ACC There are some algebraic congruent operators other [6] is than the inear one: the operator of the hyperbolic congruent (HC) sequence; [7] is the quadratic congruent [8] is the operator of (QC) operator; and the Costas sequence, and alpha is a primitive root of . According to the operators, the ACC could be divided into three categories, and each category has three types. There are a total of nine types of ACCs, whose performances, as well as capacities, are given in Table II, respectively. Table II shows several points. 1) FH code correlation constants are related to the operators: they are both 1 for linear congruent operator and 2 for hyperbolic or quadratic ones. 2) For TS codes with hyperbolic or quadratic congruent operators, correlation constants are twice that of the relative FH codes, since one hit in FH codes correlation corresponds to 2 hits of TS codes [6]. 3) The correlation constants of parallel TS codes are times . those of TSs, but their weight is increased to 4) The FH-TS codes capacity increased from , as if in FH or . TS codes, to 5) The TS code with linear congruent operators is just the prime sequence; because of its high sidelobe of auto-correlation, it is seldom used in asynchronous CDMA systems.

Fig. 1.

Illustration of FH, TS, and FH-TS codes.

TABLE II COMPARISON OF DIFFERENT ACCS PERFORMANCES AND CAPACITIES

III. CROSS-PROPERTIES BETWEEN THREE TYPES OF ACCS These three types of different code MAIs should be considered, if they were used in one hybrid system. In this section, we will analyze the cross-correlations between FH, TS, and FH-TS codes separately. In the next section, we will calculate the probabilities to achieve those cross-correlations, and then we will give out the BER performance caused by MAI. A. Cross-Properties Between FH and FH-TS Codes In order to demonstrate the interference from FH code to FH-TS, which could be represented by the cross-correlation, Theorems 13 are provided. Theorem 1: When the linear congruent operator is used, the maximum cross-correlation between FH and FH-TS codes is 1. Proof: Considering an FH code modulated by user information of all 1s, let us look into a segment of such an FH

PU et al.: RESEARCH OF ALGEBRA CONGRUENT CODES USED IN 2-D OCDMA SYSTEM

2559

modulated signal that is as long as an FH-TS code. The mathematical expression is shown as (4) time It is a long sequence occupying wavelengths and slots. The cross-correlation between this sequence and the . FH-TS one is determined by operator Supposing is the relative time shift between FH (shown in . Here, we do (4)) and FH-TS codes (shown in (3)), not consider the frequency shift. The nonzero cross-correlation happened in position s, which is constrained by (5) where represents modulo- congruent to. For the linear with . We then get operator, we substitute

words, the operator used in FH-TS has the same value as that of parallel TSs. The number of such FH-TS , parallel codes is . Since prime sequence TS and FH-TS codes could not be used simultaneously while linear congruent operators with the same index are used. There will be no such limitation when the other two algebraic congruent operators are used. when FH-TS codes use dif2) The cross-correlation is ferent TS sequences to that of parallel TSs. In other used in FH-TS has a different words, the operator value to that of parallel TSs. The number of such , and it is of the total. FH-TS codes is IV. THE BER PERFORMANCE OF HYBRID SYSTEM In a hybrid system with FH, parallel TS and FH-TS codes as address sequences simultaneously, BER performance is determined by the hits between the interference and the intended codes. The maximum hit numbers between each ACC was provided in the previous section. In this section, the maximum probability to achieve those hit numbers will be calculated by introducing Theorems 46 and their proofs. According to those hit probabilities, the contribution of different types of interferences to hybrid system BER performance could be derived. At last, we will compare the performance figures of each type of interference. Theorem 4: To a linear operator, the one-hit probability , while the between FH-TS and FH codes is smaller than two-hit probability is zero. there is one hit between FH-TS and FH Proof: codes (8) where 1/2 comes from the assumption that user datas 0s and1s are with equal probabilities. To a given code, , , , has only one solution (see and are constant, and each there are two hits between proof of Theorem 1), so . If we take the linear operator into FH-TS and FH codes (8), then we get

(6) Equation (6) shows that, when parameter , , , and are given, has only one solution in the position where satisfies (6). Therefore, Thereom 1 is proved. Theorem 2: When the quadratic congruent operator is used, the maximum cross-correlation between FH and FH-TS codes is 2. into (5), and then we get Proof: Take

(7) The two equations of (7) could be identical when a certain was chosen. They are second-order equations in field , and hence, by Lagranges theorem [9], there can be at most two incongruent solutions in the field; consequently, the quadratic operator can have at most two hits for any time shift. Theorem 3: When the hyperbolic congruent operator is used, the maximum of cross-correlation between FH and FH-TS codes is 1. Proof: This is similar to Theorem 1s proof. B. Cross-Properties Between Parallel TS and FH-TS Codes Identical TS codes could be used simultaneously on all wavelengths to provide users with high QoS service. We call it parallel TS codes. Parallel TS codes are 2-D codes with , , , ), where and are properties of ( (auto)cross-correlation constants of FH codes. Their values have been discussed in Section II-A. Obviously, there are two values for the cross-correlation between parallel TS and FH-TS codes, as follows. when FH-TS codes use the 1) The cross-correlation is same TS sequences as that of parallel TSs. In other

(9) , by Lagranges theorem, there are at most two among positions, which could satisfy (9), so . Theorem 3 is proved. Theorem 5: To a hyperbolic congruent operator, the one-hit , probability between FH-TS and FH codes is smaller than while the two-hit probability is zero. into Proof: If we take hyperbolic operator (5), then we have

(10)

2560

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 21, NO. 11, NOVEMBER 2003

Both equations of (10) should be satisfied; therefore, and is shown as follows:

To a quadratic congruent operator, (see Theorem 6). Its BERs upper-bound is

and

(11)

Since second-order equations in field have at most two in. congruent solutions, then Theorem 6: To a quadratic congruent operator, the one-hit probability between FH-TS and FH codes is smaller than , while the two-hit probability is smaller than . Proof: If we take quadratic operator into (5), then we get the two equations of (7). Two-hit happens when two equations degenerate to one second-order equation; therefore

(13) To a hyperbolic congruent operator, 5). Its BERs upper bound is (see Theorem

(14) , BER could be approxWhen is big enough and imated by Gaussian distribution [11]. The Gaussian approxi, , and mate BER of these three operators are , which are shown as follows (the details are shown in Appendix I):

and

first-order equation first-order equation two equations of (7) have a similar solution

where you obtain the equations at the bottom of the page. . Therefore, The interferences in hybrid system are divided into three types by different codes. We will show their contributes to BER one by one. A. Type A. Several FH or FH-TS Users Interference to a FH-TS User When an FH-TS codes user access into the networks together users whose address sequences are either FH or with other users will interference the former FH-TS codes, the later users correct receiving, hence cause BER. The BER caused by interference is determined by hit probabilities and , and its upper bounds are given by [10, eq.(7)] or [10, eq.(23)]. (see Theorem 4). To a linear congruent operator, Therefore, the upper bound of BER is

(15)

(16)

where

(17)

B. Type B. Several FH-TS Users Interference To an FH User of FH-TSs, the probability of Since FH codes length is of that FH-TSs probability being FH being hit by FH-TS is hit by FH, where 2 comes from the uncertainty of user data in , , and , and (8). We have BER upper bounds , , which is Gaussian approximate BER shown in (18)(20) at the bottom of the next page.

(12)

PU et al.: RESEARCH OF ALGEBRA CONGRUENT CODES USED IN 2-D OCDMA SYSTEM

2561

Their Gaussian approximations are

(21)

(22)

(23)

Fig. 2 shows that, to parallel TS codes users, the hybrid system is an error-free channel, without considering the receivers shot and thermo noise, the intensity noise of the laser source, and the chromatic dispersion of the fiber channel. In addition, type Bs BER is smaller than type As, which means that the FH codes anti-interference ability is stronger than the of the FH-TS codes, and such ability of parallel TS codes is the stronger among those three codes in the hybrid system; therefore, parallel TS codes users could enjoy high QoS. In Fig. 3, different congruent operators for type A are compared. It told us that linear operators BER is always better than hyperbolic, and hyperbolic is always better than quadratic, while type A interference was considered. In Fig. 4, different congruent operators for type B are compared. Fig. 4 shows us that linear operators BER is always better than hyperbolic, and hyperbolic is always better than quadratic, while type B interference was considered. V. CONCLUSION Until now, a hybrid 2-D OCDMA system with different bit rates and different QoS is given, in which three categories of algebraic congruent operators (linear, quadratic, and hyperbolic) are used to offer three types of 2-D code sequences (FH, TS, and FH-TS). The bit rate of FH codes is the times the other two, and the anti-interference ability of parallel TS codes is much better than the other two. By analyzing BER performance, we could see that in FH-TS codes 2-D OCDMA system, adding a small number of FH or parallel TS users would not greatly affect the original system, only if these three types of users should use the same congruent operator. This system is very fit for the case that a small number of users need high bit rate or high QoS, while a great number of users do not have this need.

C. Type C. Several FH-TS Users Interference to a Parallel TS User The cross-property between FH-TS and parallel TS is the same as the TS sequences self-property, except for the code weight being replaced by . Therefore, we could use the conclusion about TS sequences in [6]. By modifying [11, eq. (5)], we could obtain the TS users Gaussian-approximated BER, when the linear operator is used, as follows: (24) Here, we could not give quadratic or hyperbolic operators type- BER, since the time sequences BER performance has not been discussed when nonlinear operators are used. In Fig. 2, three types of Gaussian approximate BERs are is chosen. given, where

(18)

(19) (20)

2562

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 21, NO. 11, NOVEMBER 2003

(a)

Fig. 3.

Type As BER of three kinds of operators, p = 7.

(b) Fig. 4. Type Bs BER of three kinds of operators, p = 7.

APPENDIX PROBABILITY OF ERROR ANALYSIS ASSUMING GAUSSIAN DISTRIBUTION In asynchronous and noncoherent optic-fiber system, the relationship between signal-to-noise ratio (SNR) and BER was (A1) Supposing correlations between different interference and dedicated users are random variables (RVs) with identical independent distribution (i.i.d), their means and variances are

(c) Fig. 2. BER comparison of three different types of interference. (a) Linear congruent operator was used. (b) Quadratic congruent operator was used. (c) Hyperbolic congruent operator was used.

(A2) While there was a large number of interference users , the RV that represents total interference is approximated by a Gaussian distribution; its mean and variance are (A3) A. The Case of In the case of , , , the interference RV has (A4)

Performance analysis also told us that the linear congruent operator has the best performance while considering three types interferences. However, the linear congruent operator cannot be used in parallel TS codes, since the sidelobe of the PS sequence is too high, and the whole system is an asynchronous one. Therefore, the linear operator could not be considered in occasions where high QoS is needed, and comparing with quadratic operator, the hyperbolic one could afford better performance. In the occasion where only high speed needs to be added, the linear operator is the optimal one.

PU et al.: RESEARCH OF ALGEBRA CONGRUENT CODES USED IN 2-D OCDMA SYSTEM

2563

When we calculate , If we take it into (A1), we get

If we take it into (A1), we could get

(A5) (A11) When we calculate , , we have When we calculate , we have , and

(A6) When we calculate , , we have If we take it into (A1), we get

(A12)

(A7) (A13) When we calculate , , we have REFERENCES

[1] S. Kim, M. Kang, S. Park, Y. Choi, and S. Han, Incoherent bidirectional fiber-optical code division multiple access networks, IEEE Photon. Technol. Lett., vol. 12, pp. 921923, July 2000. [2] K. Yu, J. Shin, and N. Park, Novel wavelength-time spreading optical CDMA system using arrayed-waveguide grating, presented at the Optical Fiber Communications (OFC2001), Anaheim, CA, Mar. 1722, 2001, FD6-1. [3] H. Fathallah, L. A. Rusch, and S. LaRochelle, Passive optical fast frequency-hop CDMA communication system, J. Lightwave Technol., vol. 17, pp. 397405, Mar. 1999. [4] H. Fathallah and L. A. Rusch, Robust optical FFH-CDMA communications: Coding in place of frequency and temperature controls, J. Lightwave Technol., vol. 17, pp. 12841293, Aug. 1999. [5] S. P. Wan and Y. Hu, Two-dimensional optical CDMA differential system with prime/OOC codes, IEEE Photon. Technol. Lett., vol. 13, pp. 13731375, Dec. 2001. , M. D. Hahm, and E. L. Titlebaum, Constructing and [6] S. V. Maric performance analysis of a new family of optical orthogonal codes for CDMA fiber-optic networks, IEEE Trans. Commun., vol. 43, pp. 485489, Feb.Apr. 1995. , Z. I. Kostic , and E. L. Titlebaum, A new family of optical [7] S. V. Maric code sequences for use in spread-spectrum Fiber0optic local area networks, IEEE Trans. Commun., vol. 41, pp. 12171221, Aug. 1993. and E. L. Titlebaum, A class of frequency hop codes with [8] S. V. Maric nearly ideal characteristics for use in multiple-access spread-spectrum communications and radar and sonar systems, IEEE Trans. Commun., vol. 40, pp. 14421447, Sept. 1992. [9] K. Rosen, Elementary Number Theory and Its Applications. New York: Addison-Wesley, 1988. [10] M. Azizo glu, J. A. Salehi, and Y. Li, Optical CDMA via temporal codes, IEEE Trans. Commun., vol. 40, pp. 11621170, July 1992.

(A8)

B. The Case of In the case of

, , , the interference RV has

(A9) When we calculate we have , and ,

(A10)

2564

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 21, NO. 11, NOVEMBER 2003

[11] W. C. Kwong, P. A. Perrier, and P. R. Prucnal, Performance comparison of asynchronous and synchronous code-division multiple-access techniques for fiber-optic local area networks, IEEE Trans. Commun., vol. 39, pp. 16251634, Nov. 1991.

Tao Pu (S01) was born in Nanjing, China, on May 5, 1974. He received the B.S and M.S degrees in information and communication engineering from the Institute of Communication Engineering, Nanjing, China. He is currently working toward the Ph.D degree at the same school. Since May 2002, he was worked at the Advanced technology Research Center, Shenzhen University, Guangdong, China, for his doctorate in cooperative form and engaged in the field of optical code-division multiple access.

Yu Quan Li received the B.S. degree from Peking University, Peking, China, in 1968 and the M.S. degree from Electronic Science and Technology University, Chengdu, China, in 1981. In 1981, he joined the Institute of Communication Engineering, Nanjing, China, where he is currently a Full Professor. He has worked in the realm of electromagnetic fields, microwave technology, and optical communication. Mr. Li is a Senior Member of the China Electron Society.

Shu Wen Yang was born in 1939. She graduated from the Electronic Science and Technology University, Chengdu, China, as a postgraduate student in 1965. She was a Visiting Scholar at The George Washington University, Washington, DC, from 1981 to 1982. She is a Director of the Advanced Technology Research Center, ShenZhen University, Guangdong, China. Her interests focus on optical communication and networks. Dr. Yang is a Fellow of the China Electron Society and Chair of the Shenzhen Electron Society.