1362186 100年6月10曰修正替換頁 九、發明說明: ---- 【發明所屬之技術領域】 本發明係關於一種應用於軟體無線電解碼器之特徵值快速計算方法, 特別是指一種應用於里德-索羅門解碼器之特徵值快速計算方法。 【先前技術】 請參閱圖1 ’為習知運用於里德·索羅門解碼器中特徵值計算方法 之流程步驟圖其中Γη為接收多項式以幻中χη項的係數,a為GF(256)之本質 元素,η和k為0到255的整數。運用習知技術在(255,255-2t)里德-索羅. 門碼(企業標準)的症狀計算如下,這㈣【是能更正符號錯誤的數目。 接收多項式 (1)1362186 June 10th, 100th revised replacement page IX, invention description: ---- [Technical field of invention] The present invention relates to a method for quickly calculating eigenvalues applied to a software radio decoder, in particular to an application A fast calculation method for the eigenvalues of the Reed-Solomon decoder. [Prior Art] Please refer to Figure 1 for the flow chart of the eigenvalue calculation method used in the Reed Solomon decoder. Γη is the coefficient of the receiving polynomial in the illusion, and a is GF(256). The essential elements, η and k, are integers from 0 to 255. Using the conventional technique at (255, 255-2t) Reed-Solo. The code of the door code (enterprise standard) is calculated as follows. (4) [is the number of errors that can correct the symbol. Receive polynomial (1)
⑶ 明=%)琳) 這裡的7是傳送的代號,而7是錯誤多項式如 e = ejXJl + ehxh + · · · + e.x^(3) Ming=%)琳) where 7 is the code of the transmission, and 7 is the error polynomial such as e = ejXJl + ehxh + · · · + e.x^
這個是錯誤的值及{λλ,/…,χΛ丨是錯誤位置。 e.sGF(2m) and XEGF(2m) 症狀的計算如 5, = r(al) = rQa10 + r{a11 +··· + γ„ αΗΐη~Χ) —1 S2 =r(a2) = r0a2.0 +/·ια21 +…+ /· α2·(2™-ΐ) : r" (4) S2l =r(a2,) = r0a(2〇'° +r,a(201 + + ^ 關於索羅門解碼器中特徵值計算,習知技術乃利用除法技術來達成, 即圖1中的mod 255 ’該mod 255為m〇dul〇 255 (除255取餘數)的除法指 1362186 iOO年6月10日修正替換.. 令。細d運算可以在x86平台的DIV (除法)指令來得^, 中指出’ DIV指令需要23個cpu週期時間,故速度較慢,不適合軟體無、 線電的實現》 由此可見,上述習用技術仍有諸多缺失及不足,實非一良善之設計, 而亟待加以改良。 本案發明人鑑於習用技術所触的各項缺點及不足,乃盈思加以改良 創新,並經多年苦心孤諸潛心研究後,終於成功研發完成本件應用於軟體 無線電解碼器之特徵值快速計算方法。 $ 【發明内容】 本發明之目的即在於提供—種應·軟體無線電解碼器之特徵值快速 計算方法’係_加法指令(ADD)來取代除法指令(div)的使用,進 而使軟體無_解之特雖計算缺為快速。 本發月之-人目的係在於提供一種應用於軟體無線電解碼器之特徵值 快速計算方法,由於很多微處理器(cpu)都是一個16位元暫存器由兩個 8位兀暫存器組成’故本發明將不只限於X86系統之應用。 鲁 可達成上述發明目的之應用於軟體無線電解碼器之特徵值快速計算方 法,係利用加法指令(ADD)來取代目前使用的除法指令(div),由於一 個16位元暫存器(ax)乃由一個高8位元暫存器(ah)及一個低8位元暫 存器⑻所組成,該高8位元暫存器乘以256再加上低8位元組暫存器, 即為I6位7G暫存器的值,因此10位元暫存器除以255將等於兩個8位元 暫存器相加(ax=ah+al),又由於使用除法指令所需的時間遠大於加法指令,This is the wrong value and {λλ, /..., χΛ丨 is the wrong location. e.sGF(2m) and XEGF(2m) The symptoms are calculated as 5, = r(al) = rQa10 + r{a11 +··· + γ„ αΗΐη~Χ) —1 S2 =r(a2) = r0a2. 0 +/·ια21 +...+ /· α2·(2TM-ΐ) : r" (4) S2l =r(a2,) = r0a(2〇'° +r,a(201 + + ^ About Solomon The eigenvalue calculation in the decoder, the conventional technique is achieved by the division technique, that is, the mod 255 in Fig. 1 'the mod 255 is m〇dul 255 (except 255 takes the remainder) of the division finger 1362186 iOO June 10 Correction replaces.. Order. The fine d operation can be obtained from the DIV (division) instruction of the x86 platform. It indicates that the 'DIV instruction requires 23 cpu cycle time, so the speed is slower, and it is not suitable for software no, line power implementation. It can be seen that there are still many shortcomings and deficiencies in the above-mentioned conventional techniques, which are not a good design, but need to be improved. The inventors of this case have improved and innovated in view of the shortcomings and shortcomings of the conventional technology, and have been painstaking for many years. After studying hard, I finally successfully developed a fast calculation method for the feature value of the software applied to the software radio decoder. $ [Summary] The purpose of the present invention is In order to provide a fast calculation method for the characteristic value of the software-based radio decoder, the _addition instruction (ADD) is used instead of the division instruction (div), so that the software has no _ solution, although the calculation is lacking fast. Month-People aims to provide a fast calculation method for eigenvalues applied to software radio decoders, since many microprocessors (cpu) are a 16-bit scratchpad composed of two 8-bit 兀 registers. Therefore, the present invention is not limited to the application of the X86 system. The fast calculation method for the eigenvalues applied to the software radio decoder by Luke to achieve the above object is to use the addition instruction (ADD) instead of the currently used division instruction (div), A 16-bit scratchpad (ax) consists of a high 8-bit scratchpad (ah) and a low 8-bit scratchpad (8), which is multiplied by 256 plus The lower 8-bit register is the value of the I6-bit 7G register, so dividing the 10-bit register by 255 will equal the sum of the two 8-bit registers (ax=ah+al). And because the time required to use the divide instruction is much larger than the add instruction,
S 故本發明以加法指令取代除法 速。 100年6月10日修正替換頁 指令,將可使解碼器之^徵值運算能更為快 【實施方式】 由於在Intel報告指出除法指令(DIV)需要23個cpu週期。將遠大 於加法指令0.5個CPU週期。所以可清楚得知習知技術速度較慢,不適合 軟體無線電技術的實現。 _ 本發明乃在一些條件下用加法指令(ADD)來取代除法指令, 將可以得到較好的效能。 更進一步的分析,可以將X除255運算分為以下三種狀況: 狀況 1 : if ;c<255,fl/i = 0 jc mod 255 = (0x256 + a/)mod 255 =(a/) mod 255 - =al . (5) =ah + al # 狀況 2 : if = 255,aZ = 255,M = 〇 (6) ah + al = 255 xmod 255 = 0^ah + al 狀況 UI:if 256<文<510,以=1,0“/<255 λ: mod 255 =(ah x 256 + al) mod 255 =(1 x 256 + al) mod 255 =(1 + al) mod 255 = l + al =ah + al 丄 δΟ 表1.x mod 255結果Therefore, the present invention replaces the division speed by an addition instruction. Correcting the replacement page instruction on June 10, 100 will make the decoder's value calculation faster [Embodiment] Since the Intel report indicates that the division instruction (DIV) requires 23 cpu cycles. It will be much larger than the addition instruction by 0.5 CPU cycles. Therefore, it is clear that the conventional technology is slow and is not suitable for the implementation of software radio technology. _ The present invention replaces the division instruction with an addition instruction (ADD) under some conditions, and a better performance can be obtained. For further analysis, the X divide 255 operation can be divided into the following three conditions: Condition 1: if; c < 255, fl / i = 0 jc mod 255 = (0x256 + a /) mod 255 = (a /) mod 255 - =al . (5) =ah + al # Condition 2 : if = 255,aZ = 255,M = 〇(6) ah + al = 255 xmod 255 = 0^ah + al Status UI: if 256<text<;510,=1,0"/<255 λ: mod 255 =(ah x 256 + al) mod 255 =(1 x 256 + al) mod 255 =(1 + al) mod 255 = l + al = Ah + al 丄δΟ Table 1.x mod 255 results
100年6月10曰修正替換頁 結果列在表丨。從這個表中可以注意到大部分(議㈣)的例子能 用㈣來取代。因為邊有炉255,因為_d255 = 〇“y的 關係不能保持。物免糊外,_科細“(傘。⑽”。 —近似的結果如表2所示張表,可注意那行“㈣的_⑴,, 是與x<510的“xm〇d255”相同。 表 2. (x+l) m〇d 255 結果 例 X X mod 255 ah+alf〇r(x+1) ah+al-1 for (x+l) 0 幺;c < 254 X x+l X 例子n 254 254 . 255 254 例子m ---一 255 0 1 — -. 0 例子IV ----- 256 1 2 丨— 1 例子V 256 < λ < 510 x-255 --- x-254 x-255 由於這種關係,本發明設計-個新_程步驟,如圖2所示, 說明里德-索羅門碼,假設“ GF⑽)之本f元素,錢存在一組可 去除t個錯誤符號的(255, 255-2t)里德-索羅門崎,其各項參數定義 如下: 碼字長度 255The results of the revised replacement page on June 10, 100 are listed in the table. It can be noted from this table that most of the examples (decision (iv)) can be replaced by (iv). Because there is a furnace 255, because _d255 = 〇 "y relationship can not be maintained. Objects are not paste, _ section fine" (Umbrella. (10)". - Approximate results as shown in Table 2, can pay attention to that line " (d) _(1),, is the same as x<510's "xm〇d255". Table 2. (x+l) m〇d 255 Result Example XX mod 255 ah+alf〇r(x+1) ah+al- 1 for (x+l) 0 幺;c < 254 X x+l X Example n 254 254 . 255 254 Example m --- a 255 0 1 — -. 0 Example IV ----- 256 1 2 丨— 1 Example V 256 < λ < 510 x-255 --- x-254 x-255 Due to this relationship, the present invention is designed with a new _ step, as shown in Figure 2, illustrating Reed Solomon Code, assuming the f element of "GF(10)), there is a set of (255, 255-2t) Reed-Solomonsaki that can remove t error symbols. The parameters are defined as follows: Codeword length 255
S 8 1J62186 訊息位元數目 255-2t 可更正符號錯誤的個數 t 高8位元暫存器 (ah) 低8位元暫存器 (al) 100年6月l〇日修正替換頁 Γη為接收多項式χη項的係數 η和k為〇到255的整數 其流程步驟為: 步驟 1 : = l〇g(rn)+1 + d 步驟 2 : = ί·η. +、· 一 1S 8 1J62186 Number of message bits 255-2t Number of correctable symbols error t High 8-bit register (ah) Low 8-bit register (al) 100 years June 1st correction replacement page Γη The coefficients η and k of the receiving polynomial χn are integers from 〇 to 255. The process steps are as follows: Step 1: = l〇g(rn)+1 + d Step 2: = ί·η. +,· 1
步驟 3 :〜=aM-logUJ 其中log為對數運算’ anti_log為log的反運算 -上述流程步驟中並不包含除法運算,乃由加法運算所組成,故於特徵 值運算時可具有較佳之效能,適合軟體無線電技術的實現。 習用的演算法和本發明的演算法的症狀計算流量以有Intd pemium Μ 1顧z CPU χ86㈣赚4測。·工具是Mi_請㈣歸Q聰團 隊版的軟_«。魏量測工具是Mi嶋^ s_聽_效能 探測器。本發_演算法在戦平“錄計算流量是每秒85,47〇個封 包,相較於習用的演算法快了 3.06倍的時間(每秒27,917侧包)。 本發明所提供之應用於敕體無線電解石馬器之特徵值快速計算方法,盘 其他習用技術相互比較時,更具備下列優點·· 、 1.本發明在特徵值計算上省略 的指令替代。 ^改用加法雜魏度快很多倍 1362186 100年6月10日修正替終頁 2. 本發明應用於以χ86個人電腦為基礎的軟體無電DVB-T平台 上’將可使特徵值計算速度增加數倍。 3. 本發明可延伸至以χ86個人電腦為基礎的其他軟體無線電平台 上,只要有使用里德-索羅門解碼器,例如:WiMAX上。 上列詳細說明係針對本發明之一可行實施例之具體說明,惟該實施例 並非用以限制本發明之專利範圍,凡未脫離本發明技藝精神所為之等效實 施或變更,均應包含於本案之專利範圍中。 綜上所述’本案不但在技術思想上確屬創新,並能較習用物品增進上鲁 述多項功效,應以充分符合新雜及進步性之法定發明專利要件,妥依法 提出申請’騎貴局核准本件發明專獅請案,以勵發明,至感德便。 【圖式簡單說明】 圓1為習知運用於里德-索羅門解中特徵值計算〆 方法之流程步 驟圖;以及 &圖2為本發明應用於軟體無線電解竭器之特徵值快速計算^ &方法之 流程步驟圖。 φ 【主要元件符號說明】 無元件符號Step 3: ~=aM-logUJ where log is a logarithm operation 'anti_log is the inverse of log--the above process step does not include the division operation, which is composed of addition operations, so it can have better performance in eigenvalue calculation. Suitable for the implementation of software radio technology. The conventional algorithm and the symptoms of the algorithm of the present invention calculate the traffic to have 4 counts of Intd pemium Μ 1 Gu z CPU χ 86 (four). · The tool is Mi_Please (4) Return to Q Cong Team version of the soft _«. The Wei measurement tool is the Mi嶋^ s_listing_performance detector. The present algorithm is based on the calculation of the flow rate of 85,47 packets per second, which is 3.06 times faster than the conventional algorithm (27,917 side packets per second). The fast calculation method of the eigenvalues of the carcass radio calculus horse, the other advantages of the disk are compared with each other. The following advantages are obtained: 1. The invention replaces the omitting instruction in the eigenvalue calculation. Faster times, 1362186, June 10, 100, revised for the final page 2. The present invention is applied to a software-free DVB-T platform based on the χ86 PC, which will increase the eigenvalue calculation speed by several times. It can be extended to other software radio platforms based on the χ86 PC, as long as there is a Reed-Solomon decoder, such as WiMAX. The above detailed description is specific to one of the possible embodiments of the present invention, The embodiment is not intended to limit the scope of the invention, and equivalents to the embodiments of the invention should be included in the scope of the patents of the present invention. The technical idea is indeed innovative, and it can enhance the multiple functions of the conventional articles. It should be submitted in accordance with the law in accordance with the statutory invention patents that fully meet the new and progressive nature. In order to invent the invention, to the sense of virtue. [Simplified description of the schema] Circle 1 is a flow chart of the method for calculating the eigenvalues in the Reed-Solomon solution; and & Figure 2 is the application of the invention to the software The characteristic value of the wireless electrolytic device is quickly calculated. The process step diagram of the method is φ [Description of main component symbols] No component symbol
S 10S 10