New Adder-Based RNS-to-Binary Converters for the {2^𝑛+1+1,2^𝑛+1−1,2^𝑛} Moduli Set

Abstract

We investigate Residue Number System (RNS) to binary conversion, which is an important issue concerning the utilization of RNS numbers in Digital Signal Processing (DSP) applications. We propose two new reverse converters for the moduli set {2^𝑛+1+1,2^𝑛+1−1,2^𝑛}. First, we simplify the Chinese Remainder Theorem (CRT) to obtain a reverse converter that uses mod-(2^𝑛+1−1) operations instead of mod-(2^𝑛+1+1)(2^𝑛+1−1) operations required by other state-of-the-art equivalent converters. Next, we further reduce the hardware complexity by making the resulting reverse converter architecture adder based. Two hybrid Cost-Efficient (CE) and Speed-Efficient (SE) reverse converters are proposed. These two hybrid converters are obtained by combining the best state-of-the-art converter with the newly introduced area-delay efficient scheme. The proposed hybrid CE converter outperforms the best state-of-the-art CE converter in terms of delay with similar area cost. Additionally, the proposed hybrid SE converter requires less area cost with smaller delay when compared to the best state-of-the-art equivalent SE converter.

1. Introduction

The presence of carry chains in conventional weighted number systems such as binary or decimal number systems often limits the efficiency of computer arithmetic operations. Residue Number System (RNS) is a number system having an attractive carry-free property, which has proved to be highly useful in many Digital Signal Processing (DSP) applications requiring high-speed computations [1, 2]. RNS also has the following inherent features due to its carry-free property: modularity, parallelism, and fault tolerance. In order not to offset the speed gained in RNS operations, a fast RNS-to-binary converter is required. The complexity as well as the efficiency of RNS to binary converter is determined by the moduli choice and by the conversion algorithm. Many different choices of moduli sets are available for RNS to represent the binary numbers in a certain range. Three moduli sets have been actively investigated, for example, {2^𝑛−1,2^𝑛,2^𝑛+1} [3], {2^𝑛,2^𝑛−1,2^𝑛−1−1} [4], {2𝑛+1,2𝑛−1,2𝑛} [5], and {2𝑛+2,2𝑛+1,2𝑛} [6]. The dynamic range of these moduli sets is not sufficient for applications requiring larger dynamic range. Thus, the moduli set {2^𝑛−1,2^𝑛,2^𝑛+1}, which is the most popular length three moduli set, has been enhanced to {2^𝑛+1+1,2^𝑛+1−1,2^𝑛} [7] and {2^𝑛+1−1,2^𝑛,2^𝑛−1} [8], to mention just a few. Generally speaking, RNS-to-binary conversion is either based on the Chinese Remainder Theorem (CRT) [1, 2, 6] or the Mixed Radix Conversion (MRC) [9].

In this paper, we propose two new hybrid Cost Efficient (CE) and Speed Efficient (SE) reverse converters for the moduli set {2^𝑛+1+1,2^𝑛+1−1,2^𝑛}. We obtain the two hybrid converters by combining the converters in [7] with the newly introduced area-delay efficient converters. First, we simplify the CRT to obtain a reverse converter that uses mod-(2^𝑛+1−1) operations instead of mod-(2^𝑛+1+1)(2^𝑛+1−1) operations required by the proposal in [7]. Next, we further reduce the hardware complexity by making the resulting reverse converter architecture adder based. Basically, the newly introduced scheme is made up of two Carry-Save Adders (CSAs) and (𝑛+1)-bit Carry Propagate Adder (CPA). The proposed CE converter outperforms the one in [7] in terms of delay with similar area cost. Additionally, the proposed SE converter requires less area cost and smaller conversion delay when compared to the one in [7].

The rest of the paper is organised as follows: Section 2 presents the necessary background information. In Section 3, we describe the proposed algorithm. Section 4 presents the hardware realization of the proposed algorithm, and a comparison with the state-of-the-art reverse converters is provided in Section 5. The paper is concluded in Section 6.

2. Background

RNS is defined in terms of a set of relatively prime moduli set {𝑚_𝑖}_𝑖=1,𝑘 such that gcd(𝑚_𝑖,𝑚_𝑗)=1 for 𝑖≠𝑗, where gcd means the greatest common divisor of 𝑚_𝑖 and 𝑚_𝑗, while 𝑀=∏^𝑘_𝑖=1𝑚_𝑖, is the dynamic range. The residues of a decimal number 𝑋 can be obtained as 𝑥_𝑖=|𝑋|_𝑚𝑖, thus 𝑋 can be represented in RNS as 𝑋=(𝑥₁,𝑥₂,𝑥₃,…,𝑥_𝑘), 0≤𝑥_𝑖<𝑚_𝑖. This representation is unique for any integer 𝑋∈[0,𝑀−1]. We note here that in this paper we use |𝑋|_𝑚𝑖 to denote the 𝑋 mod 𝑚_𝑖 operation.

For a moduli set {𝑚_𝑖}_𝑖=1,𝑘 with the dynamic range 𝑀=∏^𝑘_𝑖=1𝑚_𝑖, the residue number (𝑥₁,𝑥₂,𝑥₃,…,𝑥_𝑘) can be converted into the decimal number 𝑋, according to the Chinese Reminder theorem, as follows [10]:𝑋=|||||^𝑘_𝑖=1𝑀_𝑖||𝑀⁻¹_𝑖𝑥_𝑖||_𝑚𝑖|||||_𝑀,(1)

where 𝑀=∏^𝑘_𝑖=1𝑚_𝑖, 𝑀_𝑖=𝑀/𝑚_𝑖, and 𝑀⁻¹_𝑖 is the multiplicative inverse of 𝑀_𝑖 with respect to 𝑚_𝑖.

This general scheme can be actually simplified when certain moduli sets of interests like {2^𝑛+1+1,2^𝑛+1−1,2^𝑛} are utilized. For this moduli set efficient converters have been presented in [7]. In [2], the New CRT proposed in [3] is formulated as follows:𝑋=𝑥₁+𝑚₁|||||𝑤₁𝑥₁+^𝑘_𝑖=2𝑤_𝑖||𝑁⁻¹_𝑖𝑥_𝑖||_𝑚𝑖|||||_{𝑚2⋯𝑚𝑘},(2)

where 𝑘>1, 𝑤_𝑖=𝑁_𝑖/𝑚₁ and 𝑁⁻¹_𝑖 is the multiplicative inverse of 𝑁_𝑖 with respect to 𝑚_𝑖. For the moduli set {2^𝑛+1+1,2^𝑛+1−1,2^𝑛}, the following relation, based on the New CRT represented by (2) has been presented in [7]:𝑋=𝑥₁+2^𝑛||2^𝑛+2𝑥₂−𝑥₁+2^𝑛+12^𝑛+1+1𝑥₃−𝑥₂||_22𝑛+2−1.(3)

Given that the residues (𝑥₁,𝑥₂,𝑥₃) have the following binary representations:𝑥₁=𝑥_1,𝑛−1𝑥_1,𝑛−2⋯𝑥_1,1𝑥_1,0,𝑥₂=𝑥_2,𝑛+1𝑥_2,𝑛⋯𝑥_2,1𝑥_2,0,𝑥₃=𝑥_3,𝑛𝑥_3,𝑛−2⋯𝑥_3,1𝑥_3,0.(4)

Equation (3) was further simplified to obtain𝑋=𝑥₁+2^𝑛||𝑣^₁+𝑣₂₁+𝑣₃||_22𝑛+2−1,(5) where𝑣^₁=𝑥_1,𝑛−1⋯𝑥_1,1𝑥_{1,0𝑛}𝑥_2,𝑛+1⋯𝑥_2,1𝑥_{2,0𝑛+1},𝑣₂₁=𝑥_2,𝑛⋯𝑥_2,1𝑥_{2,0𝑛+1}0⋯00_{𝑛}𝑥_2,𝑛+1,𝑣₃=𝑥_3,𝑛⋯𝑥_3,1𝑥_{3,0𝑛+1}𝑥_3,𝑛⋯𝑥_3,1𝑥_{3,0𝑛+1}.(6)

Given that the moduli set {2^𝑛+1+1,2^𝑛+1−1,2^𝑛} is desirable, can we obtain a more effective reverse converter when compared to the ones in [7]? In the following section, we present two effective reverse converters for the moduli set {2^𝑛+1+1,2^𝑛+1−1,2^𝑛} by first simplifying (1).

3. Proposed Algorithm

Given the RNS number (𝑥₁,𝑥₂,𝑥₃) with respect to the moduli set {𝑚₁,𝑚₂,𝑚₃} in the form {2^𝑛+1+1,2^𝑛+1−1,2^𝑛}, the proposed algorithm computes the decimal equivalent of this RNS number based on a further simplification of the well-known traditional CRT. First, we show that the computation of the multiplicative inverses can be eliminated for this moduli set resulting into memoryless reverse converters. Next, we obtain reverse converters that utilize modulo-(2^𝑛+1−1) operations instead of modulo-(2^𝑛+1+1)(2^𝑛+1−1) operations required by the state-of-the-art equivalent converters. We further reduce the hardware complexity by obtaining adder-based reverse converters.

Theorem 1. Given the moduli set {𝑚₁,𝑚₂,𝑚₃} with 𝑚₁=2^𝑛+1+1,𝑚₂=2^𝑛+1−1,𝑚₃=2^𝑛, the following holds true: ||𝑚₁𝑚₂⁻¹||_𝑚3=2^𝑛−1,(7)||𝑚₂𝑚₃⁻¹||_𝑚1=1,(8)||𝑚₁𝑚₃⁻¹||_𝑚2=1.(9)

Proof. If it can be demonstrated that |(2^𝑛−1)×(𝑚₁𝑚₂)|_𝑚3=1, then (2^𝑛−1) is the multiplicative inverse of (𝑚₁𝑚₂) with respect to 𝑚₃. |(2^𝑛−1)×(𝑚₁𝑚₂)|_𝑚3 is given by ||2^𝑛+1+12^𝑛+1−1(2^𝑛−1)||_2𝑛=||2^2𝑛+2−1(2^𝑛−1)||_2𝑛=||2^3𝑛+2−2^𝑛−2^2𝑛+2+1||_2𝑛=|||||2^3𝑛+2||_2𝑛−||2^𝑛||_2𝑛−||2^2𝑛+2||_2𝑛+1|||_2𝑛=||0−0−0+1||_2𝑛=1,(10) thus (7) holds true.
In the same way if |1×(𝑚₂𝑚₃)|_𝑚1=1, then 1 is the multiplicative inverse of (𝑚₂𝑚₃) with respect to 𝑚₁. |1×(𝑚₂𝑚₃)|_𝑚1 is given by: |(2^𝑛+1−1)(2^𝑛)|_2𝑛+1+1=|2^2𝑛+1−2^𝑛|_2𝑛+1+1=|1|_2𝑛+1+1=1, thus (8) holds true.
Again, if |1×(𝑚₁𝑚₃)|_𝑚2=1, then 1 is the multiplicative inverse of (𝑚₁𝑚₃) with respect to 𝑚₂. |1×(𝑚₁𝑚₃)|_𝑚2 is given by ||2^𝑛+1+1(2^𝑛)||_2𝑛+1−1=||2^2𝑛+1+2^𝑛||_2𝑛+1−1=||1||_2𝑛+1−1=1,(11) thus (9) holds true.

The following important relations are used in the subsequent theorem: Given the moduli set {𝑚₁,𝑚₂,𝑚₃} with 𝑚₁=2^𝑛+1+1,𝑚₂=2^𝑛+1−1,𝑚₃=2^𝑛, the following holds true:𝑚₁=𝑚₂+2,(12)𝑚₁=2𝑚₃+1,(13)𝑚₂=2𝑚₃−1.(14)

Theorem 2. The decimal equivalent of the RNS number (𝑥₁,𝑥₂,𝑥₃) with respect to the moduli set {𝑚₁,𝑚₂,𝑚₃} in the form {2^𝑛+1+1,2^𝑛+1−1,2^𝑛} is computed as follows: 𝑋=2𝑚₃𝑥₃−𝑥₁+𝑥₃+𝑚₃𝑚₁||𝑥₁−2𝑥₃+𝑥₂||_𝑚2.(15)

Proof. Equation (1) for 𝑘=3 is given by 𝑋=|||||³_𝑖=1𝑀_𝑖||𝑀⁻¹_𝑖𝑥_𝑖||_𝑚𝑖|||||_𝑀.(16) By substituting (7), (8), and (9) into (16) we obtain the following: 𝑋=||𝑚₂𝑚₃𝑥₁+𝑚₁𝑚₃𝑥₂+𝑚₁𝑚₂𝑚₃−1𝑥₃||_𝑀,=||𝑚₂𝑚₃𝑥₁+𝑚₁𝑚₃𝑥₂+𝑀−𝑚₁𝑚₂𝑥₃||_𝑀,=||𝑚₂𝑚₃𝑥₁+𝑚₁𝑚₃𝑥₂−𝑚₁𝑚₂𝑥₃||_𝑀,(17) Substituting (12) in the above equation, we obtain 𝑋=||𝑚₃𝑚₁−2𝑥₁+𝑚₁𝑚₃𝑥₂−𝑚₁𝑚₂𝑥₃||_𝑀,=|||𝑚₁𝑚₃𝑥₁−2𝑚₃𝑥₁+𝑚₁𝑚₃𝑥₂−𝑚₁𝑚₂𝑥₃|||_𝑀,=−2𝑚₃𝑥₁+|||𝑚₁𝑚₃𝑥₁+𝑚₁𝑚₃𝑥₂−𝑚₁𝑚₂𝑥₃|||_{𝑚1𝑚2𝑚3}.(18) Equation (18) can be further simplified by using the following lemma presented in [11]: ||𝑎𝑚₁||_𝑚1𝑚2=𝑚₁|𝑎|_𝑚2.(19) Applying (19), (18) becomes 𝑋=−2𝑚₃𝑥₁+𝑚₁||𝑚₃𝑥₁+𝑚₃𝑥₂−𝑚₂𝑥₃||_𝑚2𝑚3.(20) Using (14) in (20), we obtain 𝑋=−2𝑚₃𝑥₁+𝑚₁|||𝑚₃𝑥₁+𝑚₃𝑥₂−𝑥₃2𝑚₃−1|||_𝑚2𝑚3,=−2𝑚₃𝑥₁+𝑚₁|||𝑚₃𝑥₁+𝑚₃𝑥₂−2𝑚₃𝑥₃+𝑥₃|||_𝑚2𝑚3,=−2𝑚₃𝑥₁+𝑚₁𝑥₃+𝑚₁|||𝑚₃(𝑥₁−2𝑥₃+𝑥₂)|||_𝑚2𝑚3.(21) Applying (19), (21) becomes 𝑋=−2𝑚₃𝑥₁+𝑚₁𝑥₃+𝑚₁𝑚₃|||𝑥₁−2𝑥₃+𝑥₂|||_𝑚2.(22) Using (13) in (22), we obtain 𝑋=−2𝑚₃𝑥₁+𝑥₃2𝑚₃+1+𝑚₁𝑚₃|||𝑥₁−2𝑥₃+𝑥₂|||_𝑚2,=2𝑚₃𝑥₃−𝑥₁+𝑥₃+𝑚₁𝑚₃|||𝑥₁−2𝑥₃+𝑥₂|||_𝑚2,(23) thus, (15) holds true.

We reduce the hardware complexity by further simplifying (15) using the following properties [7].

Property 1. Modulo (2^𝑠−1) multiplication of a residue number by 2^𝑡, where 𝑠 and 𝑡 are positive integers, is equivalent to 𝑡 bit circular left shifting.

Property 2. Modulo (2^𝑠−1) of a negative number is equivalent to the one’s complement of the number, which is obtained by subtracting the number from (2^𝑠−1).

Assumption 1. The hardware complexity is reduced based on the assumption that 𝑥₁<2^𝑛+1 always holds true.

Based on the given assumption, 𝑥₁, which is (𝑛+2)-bit binary number can now be represented like an (𝑛+1)-bit number. Therefore, the residues (𝑥₁,𝑥₂,𝑥₃) have binary representations as follow:𝑥₁=𝑥_1,𝑛𝑥_1,𝑛−1⋯𝑥_1,1𝑥_1,0,𝑥₂=𝑥_2,𝑛𝑥_2,𝑛−1⋯𝑥_2,1𝑥_2,0,𝑥₃=𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,1𝑥_3,0.(24)

Then (15) is further simplified using the following theorem.

Theorem 3. Provided that 𝑥₁<2^𝑛+1 holds true, the binary equivalent of the RNS number (𝑥₁,𝑥₂,𝑥₃) with respect to the moduli set {𝑚₁,𝑚₂,𝑚₃} in the form {2^𝑛+1+1,2^𝑛+1−1,2^𝑛} is computed as follows: 𝑋=𝐴₀+2^2𝑛+1𝐴₁+2^𝑛𝐴₁,(25) where 𝐴₁=||𝑥₁+𝑥₂+𝑢₃||_2𝑛+1−1,𝐴₀=𝑢₁+𝑢₂,𝑢₁=𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,0_{𝑛}0𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,0_{𝑛},𝑢₂=𝑥_1,𝑛+1𝑥_1,𝑛⋯𝑥_1,011⋯1_{2𝑛+2},𝑢₃=𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,01_{𝑛+1}.(26)

Proof. We need to show that (15) can be presented as (25), and that the values of 𝐴₀,𝐴₁,{𝑢_𝑖}_𝑖=1,3 are valid. It should be noted from (15) that 2𝑚₃(𝑥₃−𝑥₁)+𝑥₃=2^𝑛+1𝑥₃+𝑥₃−2^𝑛+1𝑥₁=𝑢₁+𝑢₂ and 𝑢₁ can be represented as 𝑢₁=2𝑚₃𝑥₃+𝑥₃=2^𝑛+1𝑥₃+𝑥₃=2^𝑛+1𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,1𝑥_3,0_{𝑛}+𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,1𝑥_3,0_{𝑛}=𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,1𝑥_3,0_{𝑛}00⋯0_{𝑛+1}+𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,1𝑥_3,0_{𝑛}=𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,00𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,0_{2𝑛+1}.(27)
Also, we represent 𝑢₂ as 𝑢₂=−2^𝑛+1𝑥₁=−2^𝑛+1𝑥_1,𝑛+1𝑥_1,𝑛⋯𝑥_1,0_{𝑛+1}=−⎛⎜⎜⎝𝑥_1,𝑛+1𝑥_1,𝑛⋯𝑥_1,0_{𝑛+1}(00⋯0)_{𝑛+1}⎞⎟⎟⎠=⎛⎜⎜⎝𝑥_1,𝑛+1𝑥_1,𝑛⋯𝑥_1,011⋯1_{2𝑛+2}⎞⎟⎟⎠.(28) Next, we convert each term in |𝑥₁+𝑥₂+𝑢₃|_2𝑛+1−1 to an (𝑛+1) bit binary number. No manipulation is required for 𝑥₁ since it is treated like an (𝑛+1)-bit number and also 𝑥₂ is already an (𝑛+1)-bit number. The only term to be manipulated is −2𝑥₃ and this is carried out as follows: 𝑢₃=−2𝑥₃=−2𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,0_{𝑛}=−𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,00_{𝑛+1}=𝑥_3,𝑛−1𝑥_3,𝑛−2⋯𝑥_3,01_{𝑛+1}.(29)

However, if the condition 𝑥₁<2^𝑛+1 does not hold true, the converter in [7], represented by (5), is utilized.

4. Hardware Realization

The hardware implementations of the proposed reverse converters are based on the combination of (25) and (5). We propose two converters based on this approach, termed “hybrid approach”. The hardware realizations of the proposed schemes are depicted by Figures 2 and 3. In Figure 2 (the proposed hybrid cost-efficient converter), 𝑥₁ is first input into the system. Given that the Most Significant Bit (MSB) of 𝑥₁ is 𝑥_1,𝑛, then 𝑥_1,𝑛 is compared with “1”. If 𝑥_1,𝑛=1, (5) is utilized just as outlined in [7] otherwise the hardware realization follows (25). It should be noted that the overhead for comparison is approximately zero as only the MSB of 𝑥₁ needs to be compared with a 1. The condition 𝑥_1,𝑛=1 holds true only in very few cases. The probability of occurrence of this condition is denoted by 𝑝, and it can be seen from Figure 1 that 𝑝 approaches zero as 𝑛 increases. This is fully explained in the next section. However, if 𝑥_1,𝑛=0 (which occurs most of the time) holds true, the hardware realization is as follows. The operands 𝑥₁, 𝑥₂, and 𝑢₃ in (25) are added using CSA1 producing 𝑠₁ and 𝑐₁, which are in turn added using a one’s complement adder (this is equivalent to a CPA with End Around Carry (EAC)). Suppose that 𝐵₁ and 𝐵₂ are, respectively, used to store the results of the 2𝑛+1 and 𝑛-bit right shifting of the one’s complement adder. Since 𝑢₁ is a 2𝑛+1 bit number, it can be concatenated with 𝐵₁ with no computational hardware. The second operand 𝐵₂ is a (2𝑛+1)-bit number with 𝑛-bit of zeros. 𝐵₂ must be converted to a (3𝑛+2)-bit number by appending (𝑛+1)-bit of zeros to its MSB part. The third operand 𝑢₂, which is a (2𝑛+2)-bit number, is also made a (3𝑛+2)-bit number by appending 𝑛-bit of ones to its MSB part. 𝐵₁, 𝐵₂, and 𝑢₂ are all now (3𝑛+2)-bit numbers and are to be added using CSA2 yielding 𝑠₂ and 𝑐₂. It should be noted that (2𝑛+1) bits of the Full Adders (FAs) in CSA2 are reduced to Half Adders (HAs). The final result is supposed to be obtained by a CPA but the final result (obtained by means of simulation) is always the same as inverting 𝑠₂. Thus, the final CPA is eliminated. On the other hand, Figure 2 depicts the hardware realization of the proposed hybrid speed efficient converter. Just like the proposed hybrid CE converter, hybrid SE converter is also made up of the same level of CSAs, but the only difference is that two CPAs are utilized in parallel instead of the 1’s complement adders in Figure 2. Consequently, the conversion time is significantly reduced.

Figure 1 

Probability of occurrence of 𝑥1,𝑛=1 versus 𝑛.

Figure 2 

Proposed hybrid cost-efficient RNS to binary converter.

Figure 3 

Proposed hybrid speed-efficient RNS to binary converter.

Design Example
Given the moduli set {2^𝑛+1+1,2^𝑛+1−1,2^𝑛}, where 𝑛=3, convert the residue number (𝑥₁,𝑥₂,𝑥₃)=(2,4,3) to binary.
In order to obtain the required binary equivalent, (25) is applied as given in Table 1 where CSA1 represents the first CSA in Theorem 3. The CPA then adds 𝑠₁ and 𝑐₁ producing 𝐶_𝑟. We then obtain 𝐵₁, 𝐵₂, and 𝑢₂ as already expalined above. CSA2 is used to add 𝐵₁, 𝐵₂, and 𝑢₂ producing 𝑠₂ and 𝑐₂. It can be seen from Table 1 that 𝑅_𝑓=−𝑠₂. Thus, the final CPA can be eliminated, and consequently the conversion delay and the area cost are significantly reduced.

Table 1 

Design example.

5. Performance Evaluation

The performances of the proposed converters are evaluated in terms of area cost and conversion delay. Two efficient reverse converters have been proposed. The performance comparisons of these converters and the best state-of-the-art converter in [7] are presented in Tables 2, 3, and 4. From Table 2, it can be seen that the proposed hybrid CE outperforms the CE in [7] in terms of delay with slightly lesser or similar area cost (whenever 𝑥_1,𝑛=0, the same result otherwise). With the same condition, the proposed hybrid SE converter outperforms the SE converter in [7] in terms of both speed and area. Table 4 shows the occurrence probability and other associated parameters. We note here that in Table 4, the following notations are utilized: 𝑛 is an integer value that determines various dynamic range requirements, 𝑐 is the number of times 𝑥_1,𝑛=1 occurs within the dynamic range 𝑀 of the system, 𝑝 is the probability of occurrence of the condition 𝑥_1,𝑛=1 within the dynamic range of the system (𝑝 is computed by 𝑝=𝑐/𝑀), hybrid CE stands for the delay of the hybrid CE converter, CE [7] stands for the delay of the CE converter in [7], and hybrid SE stands for the delay of the hybrid SE converter while SE [7] stands for the delay of the SE converter in [7]. As shown in Table 4, 𝑝 reduces as the dynamic range increases (i.e., as 𝑛 increases). For example, when 𝑛=3, 𝑝=0.058824 whereas 𝑝=0.000244 when 𝑛=10. This implies, as can also be deduced from Figure 1, that the occurrence probability approaches zero as 𝑛 continues to grow. The delays of hybrid CE and SE converters and that of the CE and SE converters in [7] are depicted by Figure 4. It can be easily seen from Figure 4 that the rate of growth of the conversion delay is comparably very small in the hybrid SE. Another interesting thing to note here is that the proposed CE converter and the SE converter in [7] have nearly equal conversion delay.

Table 2 

Area-delay comparison.

Table 3 

Synthesized results: area-delay comparison.

Table 4 

Simulation results: showing the occurrence probability and other associated parameters.

Figure 4 

Delay comparison: hybrid converters versus converters [7].

Additionally, the proposed hybrid SE converter (whenever 𝑥_1,𝑛=0) and the SE converter in [7] are implemented using Xilinx92i FPGA technology for varoius dynamic range requirements (different values of 𝑛). The target technology is Xillinx (Xa3s200-4vqg100) FPGA. The performance is evaluated in terms of area (measured in terms of the number of slices) and delay (represents the total gate delay, which is measured in nanoseconds). Table 3 shows the synthesized results for various values of 𝑛, which show the superiority of our scheme over the one in [7]. Consequently, the RNS-to-binary converters proposed in this paper are better than the ones in [7].

6. Conclusions

In this paper, we proposed two new reverse converters for the moduli set {2^𝑛+1+1,2^𝑛+1−1,2^𝑛}. First, we simplified the traditional CRT to obtain a reverse converter that uses mod-(2^𝑛+1−1) operations instead of mod-(2^𝑛+1+1)(2^𝑛+1−1) operations required by the proposal in [7]. Next, we further reduced the hardware complexity by making the resulting reverse converter architecture adder based. We proposed two hybrid CE and SE converters. In each of the schemes, the converter in [7] is integrated into a newly proposed area-delay efficient scheme. The path to be followed depends on whether 𝑥_1,𝑛=1 (which occurs only in few cases) or 𝑥_1,𝑛=0. In terms of delay, the two proposed hybrid CE and SE converters require (2𝑛(𝑝+1)+𝑝+4)𝑡_FA and (𝑛(𝑝+1)+3)𝑡_FA+𝑡_MUX, respectively, while the CE and SE converters in [7], respectively, require (4𝑛+5)𝑡_FA and (2𝑛+3)𝑡_FA+𝑡_MUX, where 𝑡_FA denotes the delay of one full adder and 𝑡_MUX that of a multiplexer, and 𝑝 is the occurrence probability of 𝑥_1,𝑛=1. The proposed hybrid CE converter outperforms the one in [7] in terms of delay with slightly higher or similar area cost. Additionally, with smaller delay, the proposed hybrid SE converter also requires less area cost when compared to the one in [7].