** Wireless Engineering and Technology ** Vol. 4 No. 4 (2013) , Article ID: 38683 , 16 pages DOI:10.4236/wet.2013.44029

Parallel Algorithms for Residue Scaling and Error Correction in Residue Arithmetic

^{ }^{ }^{ }^{}

Department of Electrical Engineering, National Tsing Hua University, Hsinchu City, Chinese Taipei.

Email: hylo@ee.nthu.edu.tw

Copyright © 2013 Hao-Yung Lo, Ting-Wei Lin. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Received August 3^{rd}, 2013; revised September 11^{th}, 2013; accepted September 30^{th}, 2013

**Keywords:** Chinese Remainder Theorem (CRT); Error Correction; Error Detection; Parallel Residue Scaling; Residue Number Systems (RNS); Target Race Distance (TRD); Target Residue-Digit Difference

ABSTRACT

In this paper, we present two new algorithms in residue number systems for scaling and error correction. The first algorithm is the Cyclic Property of Residue-Digit Difference (CPRDD). It is used to speed up the residue multiple error correction due to its parallel processes. The second is called the Target Race Distance (TRD). It is used to speed up residue scaling. Both of these two algorithms are used without the need for Mixed Radix Conversion (MRC) or Chinese Residue Theorem (CRT) techniques, which are time consuming and require hardware complexity. Furthermore, the residue scaling can be performed in parallel for any combination of moduli set members without using lookup tables.

1. Introduction

Because the residue number system (RNS) operations on each residue digit are independent and carry free property of addition between digits, they can be used in highspeed computations such as addition, subtraction and multiplication. To increase the reliability of these operations, a number of redundant moduli were added to the original RNS moduli [RRNS]. This will also allow the RNS system the capability of error detection and correction. The earliest works on error detection and correction were reported by several authors [1-12]. Waston and Hasting [1,2] proposed the single residue digit error correction. Yau and Liu [3] suggested a modification with the table lookups using the method above. Mandelbaum [4-6] proposed correction of the AN code. Ramachandran [7] proposed single residue error correction. Lenkins and Altman [8-10] applied the concept of modulus projection to design an error checker. Etzel and Jenkins [11] used RRNS for error detection and correction in digital filters. In [12-16] an algorithm for scaling and a residue digital error correction based on mixed radix conversion (MRC) was proposed. Recently Katti [17] has presented a residue arithmetic error correction scheme using a moduli set with common factors, i.e. the moduli in a RNS need not have a pairwise relative prime.

In this study, we developed two new algorithms without using MRD (Mixed-radix digit) or CRT (Chinese remained Theorem) for speeding-up the scaling processes and simplifying the error detection and correction in RNS. The first algorithm is used for these purposes, through the residue digit difference cyclic property (CPRDD) within the range of, where with r additional moduli. The moduli are called the nonredundant moduli; are the redundant moduli. The interval, , is called the legitimate range, where, and the interval,

, is the illegitimate range, where

, and is the total range. This paper is organized as follows: Section II will describe the scheme the cyclic property of residue digit difference (CPRDD). Section III describes the Target Race Distance (TRD) algorithm and followed by some examples. Section IV discusses residue scaling and error correction using the TRD and CPRDD algorithms. Finally, the conclusion is given in section V.

2. Error Detection and Correction Using Residue Digit Difference Cyclic Property

Any residue digit x_{i }representation in moduli set

has its cyclic length with respect to its module number. For example, if the moduli set is (4, 5, 7, 9), then the cyclic lengths of any residue digits

are 4, 5, 7 and 9, respectively. Since these cyclic lengths are not equal, they are very difficult to use as tools for error detection and correction. Actually, there exists the property of common (uniform) cyclic length in RNS between residue digital-differences (RDD). Consider three moduli set. The residue representations and their corresponding digit-differences are shown in Table 1 and defined as the difference in value between two digits, where s are all modulo to positive values with respect to if the cycle length of is assigned.

Note that the residue digit-differences in Table 1 are obtained from if, and from if. This difference of

or in values may be positive or negative, depending upon or and

or respectively All negative values must be modulo to positive values. For example, on starred row 28, as shown in Table 1, the digit difference in value for and is

. It results in

From the cyclic property of residue-digit difference (CPRDD) in RNS, we now have the following theorem.

Theorem 1. For a moduli set

and residue representation for in RNS, there exists a cyclic property in differences between two residue digits,

or. The cyclic length can be assigned, either to or, depending upon modulo operation with respect to or.

Proof: Consider the case respective to, the residue-digit difference (RDD) between two digits in

can be in general expressed by the equation

(2-1)

where

and are integers.

For simplicity, we only consider the case of and assume, and the case of can be obtained in a similar way.

The related theorem and algorithm are described as follows.

1) In cycle 0, (the initial cycle), we have

with,

As

Table 1. Cyclic property of Residue Digit Difference.

with, we have with m_{j}’s 0s in cycle 0, where means the largest integer less than or equal to x.

Thus, the RDD has m_{j}’s “0” in the initial cycle for each modulus, i.e., in cycle 0, for all.

2) Next consider each modulusSince andthen

where

For RDD = 0 (for cycles)

then

with m_{j}’s 0 s.

For RDD = 1 (not necessary in cycle 1),

with m_{j}’s 1 s.

…

For RDD = m_{i} − 1

with s.

Corollary 1. From the above theorem, we can immediately obtain that each cycle in the residue-digit difference of x will start at location 0, and end at location

.

Corollary 2. It is easily shown that there exists m_{i} number of cycles with respect to the cyclic length of.

Proof. Since the residue-digit difference of

representation is pair-wise, the legitimate range of this pair-wise is, (from 0 through). From corollary 1, the cyclic length is. Thus the number of cycles within this cyclic length for is, and for.

Theorem 2. The algorithm of theorem 1 and its corollaries can be extended to two or more pair-wise residuedigit differences.

Proof: consider a three moduli set, we have two pairwise moduli sets, whose RDD (Residue Digital Difference) is

where is again the referenced module.

Follow the same procedure as step (2) as above.

Assume, and also pair-wise numbers

and

.

1) For

thus

“0” r_{2}’s “”

This shows that has also “0”s in cycle 0 of. The cyclic length is, and the number of cycles for is.

2) For and (a constant for any RDD), if

This shows that the in any location has also in cycle i of. The number of cycles for is still. Combining these three moduli into one set, we have cyclic length (for example,

). The number of cycles for

are,

, and, respectively. As shown in Table 1, the RDD pairs of

, and

(1, 1)

In general, and

with rows and in each row.

This completes the proof.

Example 2-1.

Consider a moduli set, and its corresponding residue digits representation set is. The cyclic length is and the number of cycles for, and are

, and, respectively.

Error detection and correction:

Before the CPRDD algorithm used for error detection and correction is described, some basic terms in use must be defined.

Definition 1: Stride distance: It is the incremental or decremental distance between moduli and in absolute value from ith cycle to cycle.

For example:.

(1) Error detection Let the moduli set be

where are the nonredundant moduli and are the redundant moduli. Since the cyclic lengths of CPRDD are constant, it is thus easily found that the number of cycles on track from the starting point 0 (or other) to its target position. In turn the distance of RDD’s can also be found.

Theorem 3. The number of cycles on track (column) from any starting point (say) to its target position can be found using the equation below;

where the stride distance between moduli and and k = the number of cycles passing through from starting point to the destination, on track If, then the number of cycles are equal to the total cycles from the starting point “0” to its target position.

Proof: Since is the number of cycles from 0 to with respect to module, and is the cyclic length, thus is the total distance from the starting point to its target position. The remaining distance for on track in the th cycle must be on the same row of on track. Thus,

.

Once the RDD’s of are found, the error detection and correction for moduli can be found just by comparing the calculated cycles or RDD with the original residue representation, pair-wise so that the error module can be detected.

The procedure for error detection by using CPRDD algorithm is summarized as follows.

1) Choose two most significant (largest) moduli as the referred moduli among the n moduli, say and.

2) Find the skip distance of a cycle

.

3) Find the digit difference

.

4) Create the equation of

or

(2-2)

5) Solve for from Equation (2-2) as the

and are known. The value of must be less than or equal to

.

6) Find the corresponding distance from the starting point to.

7) Calculate from RDD_{1}, RDD_{2}, , and check the values of, and …. If these sets’ numbers are equal, then no error occurs; otherwise, error exists.

We take the similar numerical as example 2-1 to verify this algorithm. (CPRDD)

Example 2-2. Assume that a moduli set

and number X whose residue representation is. If an error occurs at, the error detection can be described as follows.

Let us begin our procedures from the

. Since

and. Then

. Solve for, and let within legitimate range

, then.

The corresponding primary distances for these two are, respectively,

Thus, the generated results of the residue representation from and are respectively

,

.

Since the calculated results of X_{1} and X_{2} are not identical, there must be errors in one of these moduli. We cannot determine which one is erroneous. To locate the module where the error exists, at least one additional (redundant) module must be used.

The procedure for error correction by using CPRDD algorithm is essential the same as the error detection. However, two additional redundant moduli and must be added for one error correction. Note that only one redundant modulus added for error detection.

1) Choose as a referred modulus.

2) Find as the same procedures of error detection steps 2-7.

3) Examine the values of. If common value exists among, then no error occurs. If there is one and only one, say that has no common value with all other, then an error exits in modulus. This completes the error correction procedures.

The following example is illustrated here to verify this algorithm.

Example 2-3. Error correction As before we can further locate and correct a single error by adding two redundant moduli, and. Let us use the same example. The moduli set

, where and are redundant moduli and, and the residue X representation,

. If a single error occurs at, e.g., and is assigned as a reference module, then, , ,and

. From CPRDD algorithm, we can find the number of cycles for these RDD’s.

,

,

,

Since the cycle length is 9, all above values must be less than. Thus we have

.

If no errors occur, all k_{ij}’s are equal, i.e.,

.

Compared to the above results with pairwise moduli, only meets this condition. There exists no such value in.

This shows that the module is faulty, therefore we can correct it as follows: since, the.

Thus

.

This completes the error correction.

Note that the above CPRDD’s for each residue-digit difference, , and can be processed in parallel. In addition, if the referenced module is assigned to the erroneous module by chance, e.g., this algorithm will fail to locate the error. In this case, there are no values that can be found to match this condition. The way to solve the problem is, of course, to assign any other moduli, e.g., or.

The hardware design for the proposed algorithm in Example 2-3 is shown in Figure 1.

3. The Target Race Distance (TRD) Scheme

The conversion or decoding technique from residue representation to X in binary is usually accomplished using the mixed-radix digit (MRD) or Chinese remained theorem (CRT). An optimal matched and parallel converter of this kind can be seen in [13]. The MRD is shown by the following expression with weighted numbers:

where, and is the mixed-radix conversion (MRC) of x.

Optimization can be obtained using this method, as the accessed table lookup time is exactly equal to the right addition time, after immediate column stage for the tree network of the adders.

Figure 1. The hardware implementation for the proposed error and correction location algorithm can be accomplished without using lookup tables.

However, time is still consumed reading a large number of lookup tables. Additional hardware complexity is required by the adder-tree networks. An algorithm called the target race distance was with a simpler structure was developed for high-speed conversion.

TRD algorithm Suppose each residue number in the has its own track, and the distance over track _{ }from 0 (starting point) to (end point) through cycles can be expressed using

.

Obviously, the primary (no multiples of m_{i}) distance of is. To obtain the X from its residue representation of, we must find a target such that traversing the same distances over tracks respectively, i.e. when the TRD distance of each target is reached, then The TRD distance of X can be found from the following theorem:

Theorem 4. Consider the simple case of two moduli sets. Its residue representation and targets are x_{1} and x_{2} respectively. Let be the primary distance of residue x_{1} from 0 to x_{1} on the track, and be the primary distance of x_{2 }from 0 to x_{2} on track. Then the TRD distance for these two residues x_{1} and x_{2} that have the same TRD distances can be obtained by the following equation.

(3-1)

In addition, k_{1} can be calculated from the equation

where m_{1} is the cyclic length of x_{1}, and k_{1} is number of cycles, all of the integers,

.

Proof: It is easy to show that the above is the common target distance of x_{1} and x_{2}, Since

Andthus is the TRD distances for both of x_{1} and x_{2}.

Corollary: It is evident that the above theorem can be extended to n moduli set and residue number. The corresponding TRD of are therefore

In addition, k_{i} can be solved from the following equations.

…

where

Note that are the targets of moduli

respectively and the is the distance that has equal track lengths, i.e.

. That is;

.

Example 3-1 Let the moduli set be

and the residue representation be. The procedures to find the TRD distance can be described as follows:

1) Find the primary distance of residue

since and

is required , thus, and

2) Repeat the procedure 1 to find the number of cycles k_{2} and k_{3} and the last TRD distances (destinations),

and.

Since

thus

and

thus

and

The final TRD distance is the common distinction of this system for targets and i.e.

. This result can be verified as follows:

, , and

Figure 2 Shows the TRD’s on tracks and respectively.

Error detection and correction by TRD algorithm A redundant residue number system with redundant moduli will allow detection of any single error [4,14]. Consider the moduli set

and the correct residue representation. Let us

Figure 2. TRD’s on track L_{1}, L_{2}, L_{3} and L_{4}.

assume that is the redundant moduli with a single error residue representation. The TRD theorem can be used to detect this error. We find that final TRD for and does not fall into the legitimate range as follows i.e.

The final TRD distance

. If we need to locate and correct this module error, another redundant module must be added. Let us assume that for this requirement in the above residue representation.

The current redundant moduli set is

and the correct residue representation is

. Let us assume that and are the redundant moduli. With a single error

. The TRD theorem can again be used to locate and correct this error. We find that final TRD’s for dose not fall in the legitimate range, but other final TRD’s for do, falls in the legitimate range:

1) TRD for and

2) TRD for and

Thus, the error is located at module m_{2} and must be corrected to.This algorithm can also be used for multiple error corrections. However, at least three redundant moduli are required. The procedures are similar.

4. Scaling with Error Correction

The above proposed algorithm used for error detection and correction has the advantage of not requiring lookup tables. No CRT (Chinese residue theorem) decoding processes are required. However, it is still time consuming and requires extensive hardware complexity for each module having multiple-value inputs to the match unit and selecting a correct one as a output. To improve this drawback, an optimal matching algorithm is proposed here for the error correction. The following two theorems will be used and an example follows.

Theorem 5. Let m_{1} and m_{2} be two relative prime numbers in RNS for module 1 and module 2 respectively. Then there must exist the relation represented by the equation, where

so that, assuming. The and k are restricted to integers.

Proof: As a first step, let. It is easily seen that and will be satisfied. Next consider. Since there are two different pair combination

and, thus the difference between and of k will always be satisfied for, where k is restricted in integers.

Theorem 6. If the values of m_{1} and m_{2} and k in the equation are known, then p_{1}

and p_{2} can always be determined from equation

or, where p_{1}p_{2} and k are within the range:

Proof: Let the difference value of be equal to d, then d will be the integers within the range between 0 and m_{2}, i.e., , or

. These two expressions show that we can always select an integer value p, within the interval between 0 and or to satisfy the conditions or

Example 4-1 Let, and. Find the minimum values of p_{1} and p_{2} respectively from the following equation :

Since and, we have

and

(4-1)

or

(4-2)

from Equation (4-1)

so

from Equation (4-2)

, so.

This result can be verified by substituting

into the above equation. Theorem 6 is very useful as shown in the following example.

In Theorem 3 of Section III, the number of cycles on track from the starting point “0” to its target position “” can be expressed by setting, i.e.

(4-3)

where is the module i stride distance referring to module j. Similarly, the number of cycles on track from the starting point ”0” to its target position “” can be expressed by setting, i.e.;

(4-4)

Since, from theorem 3, the cyclic length of the residue digits differences reference to module m_{j} is constant (uniform), then there must exist a condition,

Eliminating the above terms from Equations (4-3) and (4-4),

where, and

Example 4-2 Let the moduli set

, and the error

, the error occurs at m_{3}.

Follow the same procedures of the Example 4-1 to use this algorithm.

(4-5)

(4-6)

(4-7)

(4-8)

Eliminating and _{ }from Equation’s (4-5) and (4-6)

,

solve for from (4-5),

or

.

Check from Equation (4-5),

This shows that the error occurs at module m_{3}. From this result, we can immediately obtain. Noting that it may happen that the assigned referenced memory moduli falls coincidentally with error memory module m_{3}. In this occurrence, we cannot find the correct (integers) values of P_{1} and P_{2} within the legitimate range. It seems that this algorithm can only detect error. To complete the error correction procedure, we can simply change the referenced module to any other and follow the same procedure as before. This guarantees that the proposed algorithm in Theorem 4 will also work well in this case. The hardware structure for illustrating this algorithm is shown in Figure 3.

The proposed TRD (target Race Distance) scheme used for error correction can be used for scaling and assigning numbers in a residue number system. A redundant residue number system (RRNS) is defined as before in an RNS with r additional moduli. The moduli

, are called the nonredundant moduli, while the extra r moduli, are the redundant moduli. The interval, , is called the legitimate range where and the interval, , is the illegitimate range, where is the total range. In the RRNS, the negative numbers within the dynamic range are represented as states at the upper extreme of the total range, which is part of the illegitimate range. The positive members are mapped to the interval, if M_{k} is odd, or, if M_{k} is even. The negative numbers are mapped to the interval

Figure 3. In the block diagram using optimal matching between multiples P_{i} m and P_{k}m_{k}, the residue digits are corrected by x_{i} - x_{4} = d_{i4}.

if M_{k} is odd or

if M_{k} is even [14].

The one-to-one correspondence between the integers of the dynamic range and the states of the legitimate range in the RRNS can be established using a polarity shift. [11], The polarity shift is defined as below.

where denotes the value X after a polarity shift and

if is odd, so thata polarity shift needs to be performed prior to correcting or scaling since belongs to the legitimate range. If a single residue digit error is introduced and corresponds to modules m_{j}, then, after a polarity shift.

where is the multiplicative inverse of moduli m_{j}_{ }i.e. and The

denotes a single residue digit error and must fall within the illegitimate range , [11].

Since, and can be represented uniquely by,where

are the coefficient from the Chinese Remainder Theorem

(CRT), i.e, , where. Note that the redundant digits are zeros if no error is introduced, while at least one redundant digit is not equal to zero if a single error is introduced. Therefore, it has the same meaning that

or is used to be the entries of the error correction.

1)2)

Although the errors detection and correction described in section II have been simplified the processes due to no need of CRT conversion. It is still hardware complex and time consuming for the residue scaling operation. To improve this drawback, a direct residue-scaling algorithm can be used. It is flexible and direct to detect and prevent the errors. The flexibility means that the scaling factor can be arbitrary chosen any single module such as, i.e. not necessarily beginning from to. in order. The direct capability means no requirement for CRT extension processes for decoding or lookup tables. The following theorem (theorem 7) and example are clarified.

Theorem 7. If the scaling factor K is one of the module set and the residue digits are, respectively, then the residue digit scaled by a factor

can be obtained using the equation

(4-9).

Proof: It is easy to show that when, and Equation (4-9) is divided by on both side, we have

(4-10).

Example 4-3. For convenient comparison of the proposed TRD algorithm to other schemes such as appeared in [14], we take the same numerical example in [11]. Let the moduli set_{, }where are regular moduli and

are redundant moduli. Then

, ,

, and

. The sufficient conditions for correcting single residue digits errors are 1), or 4, , or 2,

, The maximum

, and 2)

Thus the moduli set satisfies the necessary and sufficient conditions for correcting single errors digit. Assume and a single digit error is introduced, then.

After a polarity shift,

Follow the same procedures as shown in Example 4-2. CPRDD is applied for correction without the need for using a table.

1) Assign the moduli as the reference moduli, the following residue digit references and its corresponding CPRDD equations: are obtained

2) Choose two highest digit difference as one pair for equal target race distance e.g.

. Then the true primary RDD equations are

(4-11)

And (4-12)

where and are selected so that the two RDD are equal distances.

3) Eliminating k terms in Equation’s (4-11) and (4-12) by putting

where

, then.

4) Substituting p_{1} and p_{2} into equations (4-9) and (4-10) respectively, we have, then

, and, also,

.

5) Checking other three RDD’s

The only different module residue occurs on module number at, i.e.,. The three target distances, can be from any module residue, say, (except),..

The residue representation of X is therefore,

. If a single digit error is introduced, then,.The corresponding error is therefore

After a polarity shift,

and the scaling factor to is

. The final step must use a lookup table to obtain the result, [13].

For verifying our proposed algorithm, the table of the corresponding is not required as in [13]. The processes for finding and correcting a single error based on our method are described below.

1) Find the residue digit difference to a selected module, say as before. For verifying that our proposed algorithm detects and corrects single error without using a table, the same numerical example is used to describe the procedure as follows:

,

,

Then

2) Choose two highest digit differences as one pair for equal target race distances. e.g.

and, the following two equations can be obtained:

(4-13a)

(4-13b)

3) Eliminating k terms in (4-13a) and (4-13b) by putting

then and.

4) Substituting and into Equation’s (4-13a) and (4-13b) respectively, we have, then, and, also,

Obviously, the error is located at thus

.

Furthermore, the CPRDD algorithm can be used directly and in parallel for residue scaling and error correction. Thus the process is greatly speeded up.

Example 4-4 For convenient comparison, the same numeric example as in [13] is illustrated here. Consider, and scaling factor. If an input

and a single residue digit error, corresponding toThen

After a polarity shift,

1) Dividing by after subtracting from

, this leads,

,

,

,

.

2) Dividing by after subtracting

from

Since from above only k_{3} does not match with all other’s k_{i}, i.e. and. Therefore, there occurs an error at. Once this error is detected, it is easily found and corrected from the above equations, , which in turn

and.

that

Divided by “2”,

.

Divided by “5”

;

.

The hardware structure of this example for the residue scaling is shown in Figure 4.

Actually this algorithm can be divided by any arbitrary moduli.

Example 4-5 Divided by any arbitrary moduli, say, it must subtract from X

Then

check.

This results

It can be seen from above that

which are equal each other as expected.

Example 4-6 For processing two residue scalings and error corrections in parallel, we take Example 4-4 as an illustration. Let scaling factor, i.e., the first residue scaling factor is 2 and the second one is 5 or verse versa. It is easily shown that the extended CPRDD algorithm is used and can be completed in one cycle. That is

The result is identical

, i.e.,

, and, which are identical results as shown in Example 4-4.

Example 4-7 For error correction

Figure 4. Hardware structure of the residue scaling number for Example 4-4.

the correct , and

This shows. Therefore the error correction is made by and

, which corresponds to the value in Example 4-4, in scaling factor, (dividing by “5” part).

From above results, this checks that scaling

which is within the accuracy of the residue scaling factor.

In a general case, , this time we must modify the subtraction of and from the X, before the process of the scaling. If is the scaling factor, then the subtraction must change to

, where so that

or Let us consider the following example:

Example 4-8

of moduli set

. The scaling factor

. is assumed. Then, residue

, and can be found from

.. Thus

and.

Alternatively, it could be from other module, , where

and

which has the same number to be subtracted.

From CPRDD algorithm, the scaling processes are performed as before, we then have the following results by scaling factor;

Thus, which is exactly the value and is the most closed to.

This result can be checked using sequential steps as follows:

For

Divided by 2:

30 0 811 Divided by 7:

This result of shows that the CPRDD algorithm has the capability of parallel processing operations in residue scaling and error corrections, i.e., any combination moduli scaling factors for Ks of moduli set {m_{1}, m_{2}, , m_{k}} can be performed simultaneously.

5. Conclusions

The arithmetic operations in the residue number system for addition, subtraction, and multiplication can be speeded up by using its parallel processing properties. However, some difficult operations, such as error detection and correction, must go through conversion or decoding processes from the residue representation to the regional binary number x. This is because the decoding technique is usually accomplished using the mixed-radix digit (MRD) or Chinese Remained Theorem (CRT), which are time consuming processes requiring hardware complexity. We proposed two algorithms for scaling and error correction without the need for lookup tables or increasing the encoding process.

The Cyclic property of the Residue-Digit Difference (CPRDD) algorithm can detect and correct errors from the RNS cyclic property. Any residue moduli set has a specific cycle length, which can be obtained from the individual residue number, difference, each pair, to a reference memory module m_{i}. Once the cyclic length is known, then the original value x is easily found, and in turn, the errors can be detected and corrected.

The TRD (Target Race Distance) algorithm combined with CPRDD is used for scaling and for error detection and correction. The scaling results and error correction can be directly performed by these two algorithms without using MRD or CRT. Thus, the decoding process is significantly reduced, and the hardware structure is greatly simplified. Several examples are illustrated and verified for these two algorithms.

REFERENCES

- R. W. Watson, “Error Detection and Correction and Other Residue-Interacting Operations in a Redundant Residue Number System,” University of California, Berkeley, 1965.
- R. W. Watson and C. W. Hastings, “Self-Checked Computation Using Residue Arithmetic,” Proceedings of the IEEE, Vol. 54, No. 12, 1966, pp. 1920-1931. http://dx.doi.org/10.1109/PROC.1966.5275
- S. S. S. Yau and Y. C. Liu, “Error Correction in Redundant Residue Number Systems,” IEEE Transactions on Computers, Vol. C-22, No. 1, 1973, pp. 5-11. http://dx.doi.org/10.1109/T-C.1973.223594
- D. Mandelbaum, “Error Correction in Residue Arithmetic,” IEEE Transactions on Computers, Vol. C-21, No. 6, 1972, pp. 538-545.
- F. Barsi and P. Maestrini, “Error Correcting Properties of Redundant Residue Number Systems,” IEEE Transactions on Computers, Vol. 22, No. 3, 1973, pp. 307-315. http://dx.doi.org/10.1109/T-C.1973.223711
- F. Barsi and P. Maestrini, “Error Detection and Correction by Product Codes in Residue Number Systems,” IEEE Transactions on Computers, Vol. 23, No, 9, 1974, pp. 915-924. http://dx.doi.org/10.1109/T-C.1974.224055
- V. Ramachandran, “Single Residue Error Correction in Residue Number Systems,” IEEE Transactions on Computers, Vol. C-32, No. 5, 1983, pp. 504-507. http://dx.doi.org/10.1109/TC.1983.1676264
- W. K. Lenkins and E. J. Altman, “Self-Checking Properties of Residue Number Error Checkers Based on Mixed Radix Conversion,” IEEE Transactions on Circuits and Systems, Vol. 35, No. 2, 1988, pp. 159-167. http://dx.doi.org/10.1109/31.1717
- W. K. Lenkins, “Residue Number System Error Checking Using Expanded Projection,” Electronics Letters, Vol. 18, No. 21, 1982, pp. 927-928. http://dx.doi.org/10.1049/el:19820632
- W. K. Lenkins, “The Design of Error Checkers for SelfChecking Residue Number Arithmetic,” IEEE Transactions on Computers, Vol. C-32, No. 4, 1983, pp. 388-396. http://dx.doi.org/10.1109/TC.1983.1676240
- M. H. Etzel and W. K. Jenkins, “Redundant Residue Number Systems for Error Detection and Correction in Digital Filters,” IEEE Transactions on Acoustics Speech and Signal Processing, Vol. 28, No. 10, 1980, pp. 588- 544.
- C. C. Su and H. Y. Lo, “An Algorithm for Scaling and Single Residue Error Correction in Residue Number Systems,” IEEE Transactions on Computers, Vol. 39, No. 8, 1990, pp.1053-1064. http://dx.doi.org/10.1109/12.57044
- H. Y. Lo, “An Optimal Matched and Parallel MixedRadix Converter,” Journal of Information Science and Engineering, Vol. 10, 1994, pp. 411-421.
- A. P. Shenoy and R. Kumaresan, “Fast Base Extension Using a Redundant Modus in RNS,” IEEE Transactions on Computers, Vol. 38, No. 2, 1989, pp. 152-161. http://dx.doi.org/10.1109/12.16508
- E. D. Diclaudis, G. Orlandi and F, Piazza, “A Systolic Redundant Residue Arithmetic Error Correction Circuit,” IEEE Transactions on Computers, Vol. 42, No. 4, 1993, pp. 427-433. http://dx.doi.org/10.1109/12.214689
- S. S. Wang and M. Y. Shau, “Single Residue Error correction Based on K-Term M
_{j}-Projection,” IEEE Transactions on Computers, Vol. 44, No. 1, 1995, pp. 129-131. http://dx.doi.org/10.1109/12.368003 - R. S. Katti, “A New Residue Arithmetic Error Correction Scheme,” IEEE Transactions on Computers, Vol. 45, No. 1, 1996, pp. 13-19. http://dx.doi.org/10.1109/12.481482