CDMA移动通信系统RF优化培训讲义 第一讲: CDMA原理 培训教师：王嘉华

内 容 CDMA基础原理 IS-95系统基本结构 CDMA前向信道 CDMA后向信道 CDMA中的关键技术 2

CDMA基础原理 3

C D M A ode ivision ultiple ccess 什么是CDMA? 是800 MHz 和1900 MHz PCS 运营商的选择 满足CTIA Users’ Performance Requirements 大容量 独特的编码方案提供保密通信 Wireless telephony for the public has been evolving since the years immediately following World War II. Perhaps the largest past milestone was the introduction of commercial analog systems using the AMPS standard during the mid-1980’s. Despite the limitations of their basic technology, AMPS systems have been enormously successful, peaking public interest and today penetrating in excess of 15% of population in markets where they have been deployed. AMPS uses the tried-and-true principle of “one user, one frequency” from the earliest days of radio. This approach is called Frequency Division Multiple Access (FDMA) since users are assigned onto individually-reserved frequencies. In essence, a frequency becomes the “channel” for communication. This basic concept is still the basis for broadcasting and land mobile radio. To the FDMA concept, AMPS adds the enhancement of frequency reuse. Base station coverage is limited to areas or neighborhoods, thereby allowing the same frequency to be reused simultaneously by multiple conversations scattered across a large city or region. This creates capacity for many users. However, AMPS performance is still limited by its use of analog FM modulation and a signaling scheme that requires brief interruptions in the voice audio to allow the system and handset to exchange data and commands. To recover an AMPS signal, the receiver needs only to know the frequency and the identity of a supervisory audio tone included for control purposes. 4

什么是多址技术 ? 由于在电话和无线通信系统初期，运营商力图在一条电路上同时提供尽可能多的业务。 媒介的类型 -- 举例: 双绞线 同轴电缆 Multiple Access: 多个独立的用户对一个传输媒介的同时、私有的使用 由于在电话和无线通信系统初期，运营商力图在一条电路上同时提供尽可能多的业务。 媒介的类型 -- 举例: 双绞线 同轴电缆 光纤 空中接口(无线信号) 多址技术的优点 增加容量:为更多用户提供服务 减少资金投入 降低每用户的费用 管理方便 Transmission Medium Each pair of屗 users enjoys a dedicated, private circuit through the transmission medium, unaware that the other users exist. 5

多址技术 FDMA TDMA CDMA 信道: 在传输媒介上为每个用户单独分配的,专用的一个通道 Frequency Time Power FDMA TDMA CDMA 信道: 在传输媒介上为每个用户单独分配的,专用的一个通道 物理传输媒介是一个公共资源, 可以根据建立在不同使用技术的不同标准细分为单独的信道. 这里列出了三种主要的多址技术: FDMA Frequency Division Multiple Access 每个用户使用一个不同的频率 一个信道是一个频率 TDMA Time Division Multiple Access 每个用户使用一个时间上的一个不同窗口(时隙) (“time slot”) 一个信道是在一个指定频率上的一个指定的时隙 CDMA Code Division Multiple Access 每个用户在所有的时间内使用相同的频率,通过不同的code patterns区分 一个信道是一个唯一的 (一套) code pattern(s) Following AMPS’ deployment, several new technologies have been introduced. Time Division Multiple Access (TDMA) is the technique used by the IS-54/IS-136 and the GSM standards. These technologies transmit phase modulated digital signals, and combine multiple users onto each individual radio signal using a scheme of repeating time slots. IS-54/IS-136 allows three users per 30 KHz. signal (six users likely later with a better vocoder), while GSM allows eight users per 200 KHz. signal (possibly 16 later with a better vocoder). To hear a TDMA user, the receiver must know what frequency is being used, and which time slots contain the user’s bits. Thus, users can be thought of as divided both within frequency and time, and a channel is defined as a frequency and a time slot. CDMA operates quite differently from the foregoing technologies. In CDMA, all users generally transmit simultaneously in a “herd” on a single frequency. Therefore, neither time nor frequency can be usefully exploited to recover the signal of one user. Instead, each user transmits large numbers of “chips” to represent individual bits of its message. These chips are generated using special coding patterns which are unique user-to-user. An unrelated user’s chip pattern appears to add up to nothing when interpreted using the chip pattern of the desired user. By integrating every users’ recovered chips over the entire period of one bit, the desired user’s bit value becomes apparent and all other signals and interferences add up to approximately zero in the background. 6

CDMA特点和优势 技术 特点 7

容量大 CDMA与GSM的小区容量对比： 在相同频谱利用度的情况下，CDMA的容量是GSM的5.5倍 GSM 10 MHz 50个GSM载波(10MHz/200KHz) 每载波8个时隙(包括控制信道和业务信道) 每小区3个扇区 频率复用系数(通常)=4 可配置的站型为S444 支持有效话音信道数 =48 * 8 * /4 = 96-------------->87 CDMA 10 MHz 8个CDMA载波 每载波、每扇区20个话音信道 每小区3个扇区 频率复用系数=1 支持有效话音信道数=8 * 20 * 3 = 480 去除BCCH&SDCCH 在相同频谱利用度的情况下，CDMA的容量是GSM的5.5倍 8

话音质量高 服务质量高 Excellent 5 Good 4 Fair 3 Poor 2 Bad 1 9 CDMA 8K EVRC A-to-D C O N V E R T 64kbps VOCODER 20ms Sample Codebook Pitch Filter Formant Coded Result Feedback Loop 64kb/s PCM Existing GSM CDMA 8kb/s CDMA 13kb/s GSM EFR CDMA 8K EVRC 9

覆盖大 与GSM相比： CDMA 小区覆盖随负载的变化而变化 GSM 小区在加载的情况下，覆盖保持不变 在相同覆盖条件下， 在反向链路预算上，CDMA比GSM高5dB CDMA 小区覆盖随负载的变化而变化 在无负载的情况下，小区半径是标准GSM的3倍 在每扇区20信道的负载条件下，半径是标准GSM的2倍 GSM 小区在加载的情况下，覆盖保持不变 反向链路预算比较 在相同覆盖条件下， 基站数量大大减少，节省投资 10

频率规划简单 工程设计简单，扩容方便 11 GSM： N=4 频率复用 CDMA： N=1 频率复用 1 4 3 2 3 4 CELL 1

减少由于切换产生的掉话 减低掉话，提高服务质量 CDMA 其它无线系统 小区/扇区切换采用软/更软切换 切换是先接续再中断 小区/扇区切换采用硬切换 切换是先中断再接续 容易产生掉话 CDMA 小区/扇区切换采用软/更软切换 切换是先接续再中断 服务质量高，有效减低掉话 CDMA AMPS GSM H A N D O F Cell Site B Cell Site A Cell Site B Cell Site B Cell Site A Cell Site A B R E A K MAKE 中断 减低掉话，提高服务质量 12

术语定义 CDMA信道 或 CDMA 载波 或 CDMA 载频 CDMA Forward Channel (前向信道) 由两个1.25 MHz带宽构成的双工信道, 一个用来从基站到移动台的通信 (称作前向链路 或下行链路) ;一个用来从移动台到基站的通信 (称为 反向链路 或上行链路) 在800M蜂窝系统中, 双工间隔为 45 MHz 在1900 MHz PCS中,双工间隔为 80 MHz CDMA Forward Channel (前向信道) 1.25 MHz Forward Link CDMA Reverse Channel (反向信道) the 1.25 MHz Reverse Link CDMA Code Channel (CDMA码信道) 在前向信道或反向信道中的每个独立的二进制流l 码信道通过数学的码进行区分 前向信道中的码信道: Pilot, Sync, Paging and Forward Traffic channels 后向信道中的码信道: Access and Reverse Traffic channels 45 or 80 MHz CDMA CHANNEL CDMA Reverse Channel 1.25 MHz Forward 13

CDMA: Using a New Dimension 每个用户是喧闹的人群中一个较小的声音, 叠加了一个独特的可恢复的的码 所有用户的发射功率都必须严格控制, 使这些信号到达基站时具有相同的信号强度. Since CDMA uses a chip-integrating strategy with coding to successfully discriminate between users, multiple users can share the same frequency in the same area. This gives an inherent breakthrough in capacity, but it does require that users’ signals be held to approximately comparable signal levels at the base station receiver. Another profound consequence of this strategy determines the origins of interference. In AMPS/TDMA/GSM, interference comes from relatively distant places where the same frequency is being reused. By contrast, in a CDMA system, interference comes primarily from other users of the same sector where one user competes, and from users in surrounding sectors. Power control of all users is the key to making CDMA work. Figure of Merit: Ec/Io, Eb/No (energy per chip [bit] / interference [noise] spectral density) CDMA: Ec/Io -17 to -2 dB CDMA: Eb/No ~+6 dB 14

CDMA 是一个扩频系统 通常使用的技术力图把信号压缩在尽可能小的带宽内 直接序列扩频系统使用快速的扩展序列与输入数据混合 扩展序列在接收端独立再生, 并与接收的宽带信号一起恢复原始数据 去扩展会带来实质上的增益, 增益与频谱扩展的倍数成比例 CDMA系统使用一个较大的带宽, 从而有较大的处理增益, 从而提高了容量. Spread Spectrum Payoff: Processing Gain Spread Spectrum TRADITIONAL COMMUNICATIONS SYSTEM Slow Information Sent TX Recovered RX Narrowband Signal SPREAD-SPECTRUM SYSTEM Fast Spreading Sequence Wideband Signal The Spread-Spectrum Advantage Traditional radio communication systems use signals and modulation schemes deliberately designed to occupy the minimum bandwidth possible for the type of information being carried. The philosophy is geared toward FDMA, with the paradigm “I’ll have room for more channels in my narrow band if I make the signals as narrow as possible”. This is the approach taken for AMPS, D-AMPS (IS-54/136 TDMA), and GSM. CDMA goes in an exactly opposite philosophical direction. The relatively slow, low-bandwidth user bits are spread so that they require roughly 100 times the bandwidth they would normally occupy. On the surface, this appears extravagant and wasteful for a signal carrying just one user. However, the spreading and coding introduced earlier make it possible for multiple users to occupy the same signal, and for all users to coexist on the same frequency. These two dramatic breakthroughs more than compensate for the wider required bandwidth. At the receiving end of the CDMA link, de-spreading gives a substantial gain payback proportional to the bandwidth of the signal. This processing gain is like a “rebate” -- it can be traded in for link budget advantage to increase system performance and range, or to increase the number of users, or a mix of both. In commercial IS-95/J-Std 008 CDMA, roughly 8 db. is reserved as a comparative link budget advantage over other technologies, and the rest is used for capacity improvement. 15

扩频系统原理示意(1) Power is “Spread” Over a Larger Bandwidth 16 30 KHz 1.25 MHz 30 KHz Power is “Spread” Over a Larger Bandwidth 16

扩频系统原理示意(2) 许多码信道被单独扩展, 然后加在一起,形成一个 “复合信号” 17

扩频系统原理示意(3) 使用“正确”的数学序列可以将 任一个码信道从接收到的复合 信号中抽去出来. UNWANTED POWER FROM OTHER SOURCES 使用“正确”的数学序列可以将 任一个码信道从接收到的复合 信号中抽去出来. 18

(Base Band + Spreading Sequence) 可以恢复的运算 ORIGINATING SITE DESTINATION Spreading Sequence Input Data (Base Band) Recovered Spread Data Stream (Base Band + Spreading Sequence) The diagram above shows the basic spreading process. The user provides Input Data which must be sent to the destination and recovered. A fast-rate spreading sequence is generated according to a “recipe” known at both the originating and destination ends of the link. Both ends must make this sequence in unison and apply it for signal processing. At the originating end, a modulo-2 binary adder (an XOR gate, for example) is used to add the input data to the spreading sequence. At any instant, the output of the adder is the sum of the binary values of the User Data Input and the Spreading Sequence. The addition is carried out bit by bit, without any carrying -- I.e., 0+0=1, 0+1=1, 1+0=1, and 1+1=0 with no carry. The resulting spread spectrum chip stream is transmitted to the destination. At the destination, the received chip stream is added to the spreading sequence. Chip by chip the same binary addition process occurs. The result is recovered User Input Data. The key conclusion is to notice that whatever we DO with this spreading process, we can UNDO by despreading at the destination. All we have to worry about is being sure we can make the same spreading sequence at both ends, in synchronism. Whatever we can DO, we can UNDO. 任意的数据比特流都可以与一个扩频序列叠加 在接收端, 如果可以获得原始的扩频序列,并且正确同步, 接收到信号可以被解扩 在解扩后, 原始的数据流可以被完整无缺的恢复 19

通过 CDMA “运送 和接收” FedEx Shipping Receiving 无论在运送,接收或CDMA系统中, 封装非常重要 Data Mailer Shipping Receiving 无论在运送,接收或CDMA系统中, 封装非常重要 货物被放置在嵌套的容器内, 进行保护和编址 运送时按照一定的顺序打包, 接收时要按照相反的顺序“去包装” 在CDMA系统内,容器是扩频码 If you can accept the “What we Do, we can Undo” concept from the last slide, you’re ready for full-fledged CDMA transmission. CDMA uses multiple nested (I.e., recursive) applications of this Do-Undo concept. There are actually three specific spreading sequences, applied and then removed in order. The process is much like shipping a package. Imagine: Data is recorded on a floppy disc. The floppy disk is put in a cardboard mailer for protection. The mailer is then put in an overnight letter package with an address label. The package is shipped through the cold and cruel world in the company of total strangers and other unknown packages. The recipient tears open the overnight letter package and finds the cardboard mailer inside. She rips open the cardboard mailer and finds the floppy disc. The disk is popped into the computer and the data is recovered. CDMA uses the same technique, but the role of the containers is played by spreading codes. In fact, the purposes of the spreading codes are much like containers -- they accomplish protection of their contents, and they make it possible to address the shipment to its proper destination. 20

Spread-Spectrum Chip Streams CDMA的嵌套扩展序列 Spreading Sequence A B C Input Data X Recovered X+A X+A+B X+A+B+C Spread-Spectrum Chip Streams ORIGINATING SITE DESTINATION All three spreading sequences are used in generating the CDMA signal, whether in forward-link transmission from base station to handset, or in reverse-link transmission from handset to base station. The recipe for signal generation is slightly different on the forward and reverse links, and the spreading sequences are used for different purposes in the two directions, but all three sequences are used both ways. Both handset and base station have all information necessary to manufacture the sequences as needed, and to remain in sync with each other. In following slides, we’ll see what we do with the spreading sequences and how they are used to create the code channels used by subscribers on the forward and reverse links. But first, let’s just meet the CDMA Spreading Sequences: WALSH CODES The Walsh code is the shortest of the sequences. An individual Walsh code is 64 chips long. At the CDMA chip rate, it takes only 1/19,200 of a second to send the 64 chips which make up one complete Walsh Code. There are actually 64 individual Walsh codes, numbered 0 through 63. A special property of the Walsh codes is their uniqueness. None of them correlates with any of the others -- they match in 32 places and are opposite in 32 places. This property is called “orthogonality”. CDMA 联合使用了三种不同的扩频序列形成了独特的,强壮的信道 扩频序列在发送端和接收端都容易产生 扩展序列在发送端依次扩展, 在接收端依相反的顺序解扩, 就可以恢复原始的数据流 21

CDMA的简单模型 C0(t) (Tb=nTc) d0(t)  C0(t) d’0(t) Cm-1(t) dm-1(t) 22

CDMA的简单示例——发送端 Tb d0(t) d3(t) t t Tc W1(t) c1(t) t t s0(t) s1(t) t t s(t)=s0(t)+ s1(t) t 23

CDMA的简单示例——接收端 s(t) s(t) t t c0(t) c1(t) t t Tb Tb b’0(t) =s(t)c0(t) +1 -1 -1 +1 24

我们需要多少个扩频码 (1) --区分前向码信道 Sync Pilot FW Traffic (for user #1) Paging (for user #2) (for user #3) 一个调节到特定频率的移动台,接收某一基站的某一小区的前向信道信号 这个前向信道携带了最大可有64个“前向码信道”组成的复合信号 这些码信道中有一部分是“业务信道”,其他是使CDMA系统正常运作需要的 开销信道 一个由64个数学码组成的码集被用来区分64个可能的前向码信道, 这64个码信道可能包含在一个前向CDMA信道中 码集中的数学码被称为 “Walsh Codes” 25

我们需要多少个扩频码 (2) --区分各个基站 A B Up to 64 Code Channels 一个移动台周围可能有若干个基站, 所有这些基站使用相同的频率发射 每个基站的每个扇区都发射最多由64个码信道组成的CDMA前向信道 一个移动台必须能够区分不同基站的不同扇区, 而且只监听一组码信道 两个二进制数字序列, 称为 I and Q Short PN Sequences (or Short PN Codes) , 被用作定义不同基站扇区的识别 在一个CDMA系统中, 这两个有 Short PN Sequences 可以512种不同的方式使用, 每一种构成了一个数学码, 用来识别特定基站的特定扇区 26

我们需要多少个扩频码 (3) --区分反向码信道 RV Traffic from M.S. #1837732008 #8764349209 #223663748 System Access Attempt by M.S. #4348769902 (on access channel #1) CDMA系统必须能够唯一地识别每一个可能与基站通信的移动台 在市场上有数量众多的移动台 一个称为 Long PN Sequence (or Long PN Code) 的二进制数字序列被用来唯一地石碑每一个可能的反向码信道 这个序列非常长, 有数以亿计的不同使用方式. 每一种使用方式构成一个数学码, 用来识别一个特定的用户(被称为 User Long Code)或一个特定的接入信道 (在本课程后面会作解释). 27

Correlation of Walsh Code #23 with Walsh Code #59 Walsh Codes 属性 WALSH CODES # ---------------------------------- 64-Chip Sequence ------------------------------------------ 0 0000000000000000000000000000000000000000000000000000000000000000 1 0101010101010101010101010101010101010101010101010101010101010101 2 0011001100110011001100110011001100110011001100110011001100110011 3 0110011001100110011001100110011001100110011001100110011001100110 4 0000111100001111000011110000111100001111000011110000111100001111 5 0101101001011010010110100101101001011010010110100101101001011010 6 0011110000111100001111000011110000111100001111000011110000111100 7 0110100101101001011010010110100101101001011010010110100101101001 8 0000000011111111000000001111111100000000111111110000000011111111 9 0101010110101010010101011010101001010101101010100101010110101010 10 0011001111001100001100111100110000110011110011000011001111001100 11 0110011010011001011001101001100101100110100110010110011010011001 12 0000111111110000000011111111000000001111111100000000111111110000 13 0101101010100101010110101010010101011010101001010101101010100101 14 0011110011000011001111001100001100111100110000110011110011000011 15 0110100110010110011010011001011001101001100101100110100110010110 16 0000000000000000111111111111111100000000000000001111111111111111 17 0101010101010101101010101010101001010101010101011010101010101010 18 0011001100110011110011001100110000110011001100111100110011001100 19 0110011001100110100110011001100101100110011001101001100110011001 20 0000111100001111111100001111000000001111000011111111000011110000 21 0101101001011010101001011010010101011010010110101010010110100101 22 0011110000111100110000111100001100111100001111001100001111000011 23 0110100101101001100101101001011001101001011010011001011010010110 24 0000000011111111111111110000000000000000111111111111111100000000 25 0101010110101010101010100101010101010101101010101010101001010101 26 0011001111001100110011000011001100110011110011001100110000110011 27 0110011010011001100110010110011001100110100110011001100101100110 28 0000111111110000111100000000111100001111111100001111000000001111 29 0101101010100101101001010101101001011010101001011010010101011010 30 0011110011000011110000110011110000111100110000111100001100111100 31 0110100110010110100101100110100101101001100101101001011001101001 32 0000000000000000000000000000000011111111111111111111111111111111 33 0101010101010101010101010101010110101010101010101010101010101010 34 0011001100110011001100110011001111001100110011001100110011001100 35 0110011001100110011001100110011010011001100110011001100110011001 36 0000111100001111000011110000111111110000111100001111000011110000 37 0101101001011010010110100101101010100101101001011010010110100101 38 0011110000111100001111000011110011000011110000111100001111000011 39 0110100101101001011010010110100110010110100101101001011010010110 40 0000000011111111000000001111111111111111000000001111111100000000 41 0101010110101010010101011010101010101010010101011010101001010101 42 0011001111001100001100111100110011001100001100111100110000110011 43 0110011010011001011001101001100110011001011001101001100101100110 44 0000111111110000000011111111000011110000000011111111000000001111 45 0101101010100101010110101010010110100101010110101010010101011010 46 0011110011000011001111001100001111000011001111001100001100111100 47 0110100110010110011010011001011010010110011010011001011001101001 48 0000000000000000111111111111111111111111111111110000000000000000 49 0101010101010101101010101010101010101010101010100101010101010101 50 0011001100110011110011001100110011001100110011000011001100110011 51 0110011001100110100110011001100110011001100110010110011001100110 52 0000111100001111111100001111000011110000111100000000111100001111 53 0101101001011010101001011010010110100101101001010101101001011010 54 0011110000111100110000111100001111000011110000110011110000111100 55 0110100101101001100101101001011010010110100101100110100101101001 56 0000000011111111111111110000000011111111000000000000000011111111 57 0101010110101010101010100101010110101010010101010101010110101010 58 0011001111001100110011000011001111001100001100110011001111001100 59 0110011010011001100110010110011010011001011001100110011010011001 60 0000111111110000111100000000111111110000000011110000111111110000 61 0101101010100101101001010101101010100101010110100101101010100101 62 0011110011000011110000110011110011000011001111000011110011000011 63 0110100110010110100101100110100110010110011010010110100110010110 64个 “魔法”序列, 每一个长度为 64 chips 一个 chip是一个二进制数字(0或1) 每一个Walsh Code都正交与其他 Walsh Codes 这意味着能够通过一个“滤波”的过程, 从一个与其他Walsh codes混合的信号中识别中, 因此能够抽取一个特定的Walsh codes. 两个同样长度的二进制串是正交的, 如果他们两个异或的结果具有相同数量的“0”和“1” The Walsh Codes (if you’re getting bored, pretend these are named in honor of the legendary musician Joe Walsh) are actually from a family known as Walsh-Hadamard sequences. Their special useful property is that they are all apparently random with respect to each other, and this randomness is actually perfect -- they are totally orthogonal. Another nice feature of the Walsh Codes is the ease with which they can be generated or stored for use in portable wireless devices. Walsh Codes are composed using a simple replication technique and can be built to bit lengths equal to any integer power of two. In IS-95 CDMA, we use 64 Walsh Codes (numbered 0 through 63), and each Walsh Code is 64 bits long. If you’re wondering what we do with the Walsh Codes and just can’t wait, they are used on both the forward and reverse CDMA links. However, their most famous and generally understood job is on the forward link, where each person talking on a specific BTS sector is assigned a personal Walsh code which is theirs alone for as long as they continue that call on that sector. The phone can pick out just this user from that BTS sector’s total signal by setting its own internal Walsh Code generator to the same Walsh Code. We’ll repeat and explain that function in a big-picture view coming up soon. EXAMPLE: Correlation of Walsh Code #23 with Walsh Code #59 #23 0110100101101001100101101001011001101001011010011001011010010110 #59 0110011010011001100110010110011010011001011001100110011010011001 XOR 0000111111110000000011111111000011110000000011111111000000001111 Correlation Results: 32 1’s, 32 0’s: Orthogonal!! 28

Walsh码的特点 Walsh码的特点： 彼此完全正交 长度为M的Walsh码组所包含的码的个数也为M， 不同的码按照它在矩阵中所在行的序号分别称为 W0、W1、……WM-1 用长度为M的Walsh码做扩频码时，扩频因子应当 是M，这样接收端可以完全恢复发送端的波形。 29

Walsh Codes的产生 W1 = 0 W2 = W4 = W2 n = Wn W1 = 1 0 0 0 0 0 1 0 1 0 0 0 1 W2 = 0 0 0 0 0 1 0 1 0 0 1 1 0 1 1 0 W4 = W2 n = Wn W1 = 1 1 1 1 0 1 1 1 1 1 0 1 0 1 1 0 0 1 0 0 1 30

CDMA QPSK Phase Modulator Using I and Q PN Sequences Short PN Code I Q 32,768 chips long 26-2/3 ms. (75 repetitions in 2 sec.) CDMA QPSK Phase Modulator Using I and Q PN Sequences I-sequence Q-sequence S cos wt sin wt chip input QPSK- modulated RF Output * *在基站测, I 和 Q 是同相的. 在移动台, Q 被延迟1/2 chip 从而避免 过零振幅点,那样需要一个线性的功率 放大器 short PN code 包含两个 PN Sequences, I 和 Q, 每一个的长度为 32,768 chips 两个PN Sequences 的产生类似,但移位寄存器的生成多项式不同 他们总在一起使用, 即用在QPSK的两相正交调制上 The Short PN Sequence is actually a pair of sequences -- the “I” sequence and the “Q” sequence -- which are generated in a pair of slightly-differently-tapped 15-bit shift registers. As you might suspect from the letters I and Q, this pair of sequences is used to inplement RF phase modulation both on the CDMA forward linkand on the CDMA reverse link. Although the Short PN sequence is used on both forward and reverse links, it plays an especially crucial role on the forward link. Each CDMA BTS sector is assigned a unique, distinguishable timing offset of the short PN sequence. This timing delay is called its “PN Offset”. Individual CDMA handsets can then focus on just the sector they want to hear by setting their own internal Short PN Sequence generators to match the timing offset of that specific sector. That sector’s energy survives and passes through that correlation test; other sectors, with their different timing delays, don’t match the phone’s short PN delay and hence appear random, averaging to zero and dropping out of view. We’ll tie all of these codes and their applications together in the big picture momentarily. 31

伪随机码——m序列 最大长度线性反馈移位寄存器序列 m序列的产生 x1 x2 xn Cn C2 C1 输出 生成多项式为本原多项式 32

PN Sequences: 产生和属性 伪噪声序列 PseudoNoise (PN) sequences 是由移位寄存器产生的 An Ordinary Shift Register Sequence repeats every N chips, where N is number of cells in register 伪噪声序列 PseudoNoise (PN) sequences 是由移位寄存器产生的 无反馈网络的移位寄存器: 形成序列的长度 =寄存器的个数 带抽头的移位寄存器产生一个长度为2N-1 chips的序列, (N为移位寄存器的个数), 序列具有下列性质: 这样的序列如果逐位比较, 是匹配的(明显地, 任何序列同它自己是匹配) 这样序列同他们本身是近似正交的, 如果在时间上没有精确对准的话. False correlation is small A Tapped, Summing Shift Register Sequence repeats every 2N-1 chips, where N is number of cells in register A Special Characteristic of Sequences Generated in Tapped Shift Registers Compared In-Step: Matches Itself Complete Correlation: All 0’s Sum: Self, in sync: Sequence: Compared Shifted: Little Correlation Practically Orthogonal: Half 1’s, Half 0’s Self, Shifted: In addition to the Walsh Codes, DSSS IS-95 CDMA uses two other types of pseudo-random spreading sequences. One is called the PN Short Code, and the other is called the PN Long Code. Both codes are generated using special tap-summed shift registers. If you’re familiar with shift registers, an ordinary one is rather boring. It holds a number of bits equal to the number of cells in the register. When the clock pulses, each bit moves one cell down the register and the last cell wraps back to the first. If you’re watching the bits pass by, you’ll see the same sequence again and again. If the register has N cells, the pattern is N bits long. It becomes more interesting (Dr. Timothy O’Leary would love it) if the register is cross-connected at several points so that the value of the last cell influences the values of several other cells in the shift register. This type of structure no longer produces a boring N-bit long pattern. Instead, it generates a self-mutating pattern that keeps changing and changing and changing with a total length of 2N bits. Thus a 15-bit tapped shift register produces a bit pattern 32K long, and a 42-bit tapped shift register produces a bit pattern about 4 x 1013 long. The remarkable property of the sequences generated in this type of register is that they appear quite random. A 64-bit or longer chunk taken from one point in the sequence appears essentially uncorrelated when compared with a 64-bit or longer chunk taken from a different point in the sequence. This property allows us to use a single sequence like this to distinguish many different users, by letting each user use the same sequence but with a different time delay. You’ll see applications shortly. 33

Sector/Cell 标识--PN OFFSET 11010010010110011010011001011011010011001011001100110011010011001011101000011001100101101001110101011000111010100010100110001010011000000000000000 10011001011011011100000010110011011101011001000011101011001010110111010101011000111010110011001011000001001110000100110011001110101000000000000000 1 2 511 I Q 64 chips 800 1600 2400 3200 4000 4800 5600 6400 7200 8000 feet chips 3 4 5 6 7 8 9 10 50400 51200 63 64 5200 52800 65 66 miles 67 62 61 60 59 58 57 PN = 0 PN = 1 34

CDMA “Short” and “Long” PN Codes CDMA系统使用三个 PN code 序列: 两个 “short” 和一个 “long” 两个 short PN codes (称为“I” 和 “Q”) 被用来在前向链路中正交调制区分不同基站或小区 这两个short codes由15位 PN code 发生器产生. 产生的字串的长度位 215 -1 bits , 系统在在最长连“0”后面插入一个“0”, 长度变为32768; 重复周期是 26.666... milliseconds (2 s重复 75次). long PN code 在前向链路中被用来扩展和数据扰码/随机化; 在反向链路中用来区分移动台. long code 是由 42-bit PN code 发生器产生. 产生的字串在没有插“0”之前长度是 242 -1 (大约 4.4万亿) bits long; 它的循环周期大约 41 天, 10 小时, 12 分钟 和19.4 秒. 这三个 CDMA PN codes 都同步于系统时间起点 (January 6, 1980 at 00:00:00 hours) SHORT PN SEQUENCE There are actually two Short Pseudo-random Noise (PN) sequences, always used together as a pair. The pair is responsible for controlling the I and Q axes during the phase modulation of the CDMA RF carrier. The Short PN sequences are generated in two 15-bit tap-summed shift registers. We normally speak of them in the singular as if there were only one. The Short PN sequence has a very unique property. Compared against itself exactly in sync, it matches perfectly, as any sequence would. But compared against itself out-of-sync by any number of chips, it appears virtually orthogonal -- no correlation! The short PN sequence requires 26-2/3 milliseconds to complete; in other words, it completes exactly 75 times in each two-second period. LONG PN SEQUENCE The Long Pseudo-random Noise (PN) sequence lives up to its name. Used at the CDMA chip rate of 1,228,800 per second, it takes more than 40 days to complete! The long PN sequence is generated in a 42-bit tap-summed shift register, masked with another 42-bit register to produce arbitrary timing shifts. The long PN sequence has the same unique “match in step, don’t correlate out-of-step” properties as the short PN sequence. In fact, short samples of the long PN sequence even appear orthogonal to other short samples of the same sequence. 35

前向码信道 PILOT: WALSH CODE 0 SYNC: WALSH CODE 32 Paging Walsh 1 Walsh 6 Walsh 11 Walsh 20 Sync Walsh 32 Walsh 42 Walsh 37 Walsh 41 Walsh 56 Walsh 60 Walsh 55 PILOT: WALSH CODE 0 导频信道是一个“信号灯”, 不包含任何数据信息. 它在系统捕获时作为时间参考, 在切换时作为测量的参考 SYNC: WALSH CODE 32 同步信道携带了系统标识信息和移动台在系统捕获期间需要使用的系统参数 PAGING: WALSH CODES 1 up to 7 可配置1~7个寻呼信道,由容量需要决. 他们携带寻呼消息, 系统参数消息, 和呼叫建立命令 TRAFFIC: any remaining WALSH codes 业务信道是指配一个单独的用户用来传输呼叫数据. 所有剩余的Walsh code 都可用, 仅受限于由于噪音影响的总容量 Walsh Code 0 is always the Pilot signal. The Pilot is formed by taking the Short PN sequence, adding it to Walsh Code 0 (which is a string of all 0’s), and so the Pilot is really a little sample of pure PN energy. The Pilot is transmitted to serve as an easy-to-measure test chipstream that mobiles can rapidly and frequently sample to see they can hear the signal of that BTS Sector. It’s used as the mobile first wakes up, and it’s used constantly at all other times. The mobile can measure the strength of one sector’s pilot in just a few tens of milliseconds. Walsh Code 32 is the Sync Channel. The Sync Channel contains a slow, easy-to-read bitstream that contains timing information for the long and short codes as well as frame timing for the CDMA signal and the ID number of the system. Walsh Code 1 is the primary Paging Channel. Mobiles listen to this when they’re idle, not in a call, to hear pages and various messages about parameters, feature notifications, and orders. It is possible configure up to 7 paging channels, but virtually all operators today use just one per sector. All the remaining Walsh Codes are up for grabs, assigned dynamically as new calls request to use the BTS sector or other existing calls request to begin soft handoff with this sector. 36

前向信道的编码过程 S BTS (1 sector) MTX BSC A Forward Channel is identified by: FEC Walsh #1 Sync Walsh #32 Walsh #0 Walsh #12 Walsh #23 Walsh #27 Walsh #44 Pilot Paging Vocoder more Trans- mitter, Sector X S I Q Short PN Code PN Offset 246 A Forward Channel is identified by: its CDMA RF carrier Frequency the unique Short Code PN Offset of the sector the unique Walsh Code of the user The signal from one sector of one cell is made unique from all other sectors and cells by exploiting the Short PN sequence. Each sector uses the short PN sequence, but at a different PN offset from every other sector in the area. To receive one sector, a handset needs only to set its internal PN generator to match the PN offset of the sector. PN offsets are assigned in 64-chip steps, so there are 32768/64=512 possible offsets, numbers 0-511. Each user talking on this sector is made unique by mixing with a personally-assigned Walsh code, and the resulting chip stream is called a forward traffic channel. All idle handsets listen for parameters, pages, and possible short messages on a public message stream called the Paging channel. The primary paging channel uses Walsh code #1, but if traffic dictates, additional Paging channels can be assigned (Walsh codes 2-7). Walsh code 32 carries a continuous stream of system information called the Sync Channel. This is used briefly by handsets as they wake up and acquire the system, mainly to get timing information. Walsh code 0 is occupied by a continuous stream of all zeros called the Pilot. This steady signal is measured by handsets as they acquire the system or decide when to hand off. There are 64 Walsh codes; Pilot, Sync, and Paging use at least three. The rest are available as forward traffic channels, subject to hardware availability and limitations due to RF noise generated by other users. That’s the complete description of one sector in the forward direction. 37

反向码信道 反向码信道有两种类型: 业务信道(TRAFFIC CHANNELS) : 每个用户在呼叫过程中使用业务信道传送话音数据 一个反向业务信道由一个用户特定的公共长码或保密长码定义 市面上有数量众多的CDMA手机, 因此就有数量众多的反向业务信道 接入信道(ACCESS CHANNELS): 未处在通话状态下的移动台使用接入信道传送registration requests, call setup requests, page responses, order responses, 和其他的信令消息 一个接入信道由与用户无关的公共 长码掩模定义 接入信道同寻呼信道是成对使用的. 一个寻呼信道最多可对应32接入信道 REG 1-800 242 4444 BTS Letting each mobile have its own world-unique Long-PN offset keeps everybody distinct. However, it also complicates the process of recognizing when a specific mobile wants to initiate contact with the system. How can the system know that a specific mobile (among billions of possible mobiles) will suddenly appear at a specific instant (out of quadrillions of possible instants) on a specific BTS sector (out of tens of thousands of sectors) and want to make a call? We escape this dilemma by letting the BTS broadcast the “recipe” for some public Long PN offsets which the BTS monitors continuously. When a mobile wants to initiate contact with the BTS, it simply uses THAT Long-PN offset for the initial transmissions. These public Long-PN offsets used for accessing the BTS are called ACCESS CHANNELS. It is possible to define and configure up to 32 reverse link access channels for each forward link paging channel. However, virtually all operators are presently enabling only one access channel per paging channel. In a busy multi-carrier system, there might be enough traffic to require multiple access channels. 38

反向信道的编码过程 MTX BSC BTS (1 sector) A Reverse Channel is identified by: Channel Element Access Channels Vocoder more Receiver, Sector X Long Code Gen User Long Code A Reverse Channel is identified by: its CDMA RF carrier Frequency the unique Long Code PN Offset of the individual handset Every handset has its own unique offset of the PN Long Code. The offset value is determined mathematically by the ESN of the handset. With 40+ days of 1,228,800 chips/second to choose from, there are billions and billions of reverse traffic channels mathematically possible. One small problem: How does a base station sector know to listen for the long code of a particular handset that just appeared out of nowhere and wants to make a call? There are billions of long codes to monitor! Solution: Associated with each Paging channel in the forward direction, there are 32 publicly-defined Long Code offsets reserved for reverse-direction public traffic such as call originations, registrations, etc. These reserved Long Codes are called Access Channels. After a handset is recognized on an Access Channel, its identity is known and it can be directed to a traffic channel where it will use its own natural long code. The number of reverse traffic channels which can be simultaneously active in a particular sector is limited mainly by available hardware (number of equipped channel cards), as well as by the level of RF interference from other users in the same and surrounding sectors. AN ANALOGY: A BTS sector operates much like a hotel. Forward direction: It has a street address (PN offset). It has numbered guest rooms (Walsh Codes). It has a lobby (Paging, Walsh code 1), restaurant (Sync, Walsh code 32), and an advertising sign out front (Pilot, Walsh code 0). Reverse direction: Each guest has a unique name (Long Code offset). 39

小结: CDMA Channels--SUMMARY FORWARD CHANNELS LONG CODE: Data Scrambling BTS WALSH CODE: Individual User SHORT PN OFFSET: Sector REVERSE CHANNELS WALSH CODES: used as symbols for robustness SHORT PN: used at 0 offset for tracking LONG CODE OFFSET: individual handset Summing up: On the Forward Link: One BTS sector is an independent little CDMA world with its own private PN offset. Each user on that sector, as well as pilot, sync, and paging channels, all have their own discrete Walsh Codes. To hear one signal from one sector, all a phone has to do is set its Short PN generator to match the sector’s PN offset, and use the Walsh Code of the user it’s trying to hear. It would also help if the phone’s internal clock matched the system clock, but that’s easy to establish and maintain, as we’ll see later. The Long PN code is used on the Forward Link, but not in its regular form at 1,228,800 chips/second. It is divided down by a factor of 64 (“decimated”) and the resulting slower signal (19,200/second) is binary-added to the symbols being transmitted, to make them appear statistically random and to help keep them private. On the Reverse Link: One Mobile transmits a signal unique in all the world -- it has its own private Long PN offset. To hear that mobile, all the BTS has to do is assign a channel element to the job of decoding the mobile. In that channel element, a little Long-PN generator is set to the timing offset the mobile is using. The mobile’s signal now matches the locally-generated PN and correlates with it, producing readable output. Other mobiles don’t match and fade into the background noise. The Walsh Codes are used as information symbols, and the Short PN sequence is used to measure propagation delay. 这三种扩频码在前向和反向链路中使用的方式不同 一个前向码信道由一个指配给用户的Walsh Code 和一个指配给扇区的特定PN offset 定义 一个反向信道由一个使用长码序列的特定偏移定义 40

频谱使用和系统容量: AMPS, D-AMPS, N-AMPS CDMA 30 10 kHz 200 kHz 1250 kHz 1 3 1 Users 8 Users 2 4 5 6 7 Typical Frequency Reuse N=7 Typical Frequency Reuse N=4 Typical Frequency Reuse N=1 Vulnerability: C/I @ 17 dB C/I @ 12-14 dB Eb/No @ 6 dB GSM 17 dB = 101.7 @ 50 14 dB = 101.4 @ 25 12 dB = 101.2 @ 16 22 Users 每一种无线技术 (AMPS, NAMPS, D-AMPS, GSM, CDMA)使用与它们自身独特的信号特定相适应的调制技术 无线系统的总业务容量主要取决于无线信号特点和射频设计(频率复用) 无线信号的抗干扰能力决定了可以容忍多少干扰, 因此也决定了使用相同频率的小区在空间上需要隔离的距离 对于一个特定的S/N要求, 信号带宽决定了多少射频信号可以“赛”进运营商取得牌照的频谱范围 41

Eb/N0 和 S/N 之间的关系 Eb = S R Signal Power Bit Rate = N0 = N W Noise Power Bandwidth X Eb N0 Signal to Noise Processing Gain E / t B / t 1,250,000 14,400 87 1.94 10 19.4 dB 9,600 130 2.11 21.1 8 Kb vocoder (Full Rate) 13 Kb vocoder 42

Þ CDMA系统的S/N 优点 10 0.6 10 -1.34 10 1.94 -13.4 dB = 10-1.34 @ 0.046 = 1 AMPS N-AMPS D-AMPS GSM CDMA Analog FM DQPSK GMSK QPSK/OQPSK 30 KHz. 10 KHz. 200 KHz. 1,250 KHz. C/I @ 17 dB C/I @ 12-14 dB Eb/No @ 6dB Tech-nology Modulation Type Channel Bandwidth Quality Indicator S/N @ 17 dB S/N @ 12 to 14 dB S/N @ –13.4 dB S/N 17 dB = 101.7 @ 50 14 dB = 101.4 @ 25 12 dB = 101.2 @ 16 -13.4 dB = 10-1.34 @ 0.046 = S N Þ 10 0.6 10 1.94 = 10 -1.34 -13.4 dB Signal to Noise Processing Gain (W/R) X Eb N0 1 22 43

(carrier/interference ratio) 其他无线系统: 降低干扰的措施 2 3 4 5 6 7 1 AMPS-TDMA-GSM 在以前的无线系统中,需要得到信号必须具有足够的强度来克服干扰的影响 AMPS, TDMA 和 GSM 依靠物理距离隔离来保证干扰幅度较小 同信道用户被保持在安全距离以外, 通过仔细的频率规划 相邻的用户和小区必须使用不同的频率来避免干扰 Figure of Merit: C/I (carrier/interference ratio) AMPS: +17 dB TDMA: +14 to 17 dB GSM: +12 to 14 dB 44

十五分钟! 先喝杯茶吧. 45

IS-95系统基本结构 46

cdma网络结构 . 47 OMC VLR OMC-R/ OMC-S BTS SMC PSTN ISDN AIN BC VM MSC BSC IWF PIWF PSPDN VAN INTER NET HLR/IN/AC 其中: AC : 鉴权中心 AIN : 智能网 BC : 计费中心 BSC : 基站控制器 BTS : 基站收发信台 HLR : 归属位置寄存器 ISDN : 综合业务数字网 IWF : 交互功能 MSC : 移动交换中心 OMC : 操作维护中心 OMC-S : 交换操作维护中心 OMC-R : 无线操作维护中心 PSTN : 公用电话交换网 PSPDN : 公用分组交换数据网 PIWF : 分组交互功能 SMC : 短消息中心 VAN : 增值网 VLR : 拜访位置寄存器 VM : 语音邮件 47

CDMA 前向信道 48

Drawn to actual scale and time alignment 基本的扩展和接扩示例: User’s Symbols Spread, Sent, Recovered 在BTS侧(发射): 输入 A: 用户数据速率 19,200 symbols/second 输入 B: Walsh Code #23 速率: 1.2288 Mcps 输出: 扩展信号 MS侧(接收): 输入 A: 接收的扩展信号 输入 B: Walsh Code #23 速率: 1.2288 Mcps 输出: 用户数据, 速率19,200 symbols/second 同,基站发出的数据一致 Drawn to actual scale and time alignment XOR Exclusive-OR Gate 1 Input A: Received Signal Input B: Spreading Code Output: User’s Original Data Input A: User’s Data Input B: Spreading Code Spread Spectrum Signal Originating BTS Destination MS via air interface 49

CDMA: 码----隐藏的“魔法” QPSK RF S 去调制后的 接收信号 接扩序列 (Locally Generated, =0) matches opposite Decision: Matches! ( = 0 ) Time Integration 1 Opposite ( =1) +10 -26 Received energy: Correlation -16 BTS S if 1 = if 0 = 1 Analog Summing Users 这张图演示了CDMA信号产生和恢复的的基本技术. 在IS-95系统中实际使用的编码技术包含了另外的处理,这里为了简化起见 , 没有列出 50

前向业务信道: 产生 51 I PN Q PN bits symbols chips CHANNEL ELEMENT 1.2288 Mcps Gain Control Baseband Filter I PN Q PN 1.2288 Mcps Walsh Function Block Interleaving R = 1/2, K = 9 Convolutional Encoding & Repetition 19.2 Ksps 19.2 Ksps Decimator Long PN Code Generator Scrambling User Address Mask (ESN-Based) 9600 bps 4800 bps 2400 bps 1200 bps or 14400 bps 7200 bps 3600 bps 1800 bps (From Vocoder) Power Bit 800 Hz M U X Symbol Puncturing (13 Kb only) 28.8 bits symbols chips CHANNEL ELEMENT 51

CDMA: 码----隐藏的“魔法” QPSK RF S 去调制后的 接收信号 接扩序列 (Locally Generated, =0) matches opposite Decision: Matches! ( = 0 ) Time Integration 1 Opposite ( =1) +10 -26 Received energy: Correlation -16 BTS S if 1 = if 0 = 1 Analog Summing Users 这张图演示了CDMA信号产生和恢复的的基本技术. 在IS-95系统中实际使用的编码技术包含了另外的处理,这里为了简化起见 , 没有列出 52

Forward Traffic Channel Generation Gain Control Baseband Filter I PN Q PN 1.2288 Mcps Walsh Function Block Interleaving R = 1/2, K = 9 Convolutional Encoding & Repetition 19.2 Ksps 19.2 Ksps Decimator Long PN Code Generator Scrambling User Address Mask (ESN-Based) 9600 bps 4800 bps 2400 bps 1200 bps or 14400 bps 7200 bps 3600 bps 1800 bps (From Vocoder) Power Bit 800 Hz M U X Symbol Puncturing (13 Kb only) 28.8 bits symbols chips CHANNEL ELEMENT 53

Frame Contents: can be a mixture of 可变速率的声码器 和多路技术 (仅在业务信道中使用) DSP QCELP VOCODER Codebook Pitch Filter Formant Coded Result Feed- back 20ms Sample Rate Set 2 Frame Sizes bits Full Rate Frame 1/2 Rate Frame 1/4 Rt. 1/8 36 72 144 288 Frame Contents: can be a mixture of Voice Signaling Secondary 声码器压缩话音,减小比特率 CDMA 使用一个可变速率的声码器 当通话时使用全速率 当通话停止时使用低速率 增加系统容量 更自然的声音效果 声音，信号，和用户第二部分的数据将混合在CDMA帧内 54

卷积编码器 输入 输出 55

(Data Bit is discarded) 速率 1/2， k=9 卷积编码器 Code Symbol Output 1 2 3 4 5 6 7 8 g c Data Bit Input (Data Bit is discarded) 通过编码器由信息比特产生的symbol与所有目前在所有在寄存器中的比特都相关 每一个信息比特对产生的多个symbol都有影响 通过这种方式的内部联系对侦察和纠正误码有帮助 以为寄存器中正1的长度 叫做卷积编码器的“约束长度” (在上图中K=9 ) 寄存器的长度越长, 纠错性能越好 在同等准确度的要求下，编码信号比非编码信号所需的功率小 这里每输入一个比特产生两个信号(速率 1/2) 56

“I”与“Q”复合 + + + + + + + + + + + + S S “I” PN Code “Q” PN Code Walsh code + Pilot Channel 在同一个区域（小区或扇区）中的前向信道与相同的I 和 Q 序列结合 为了把这小区与其他与其相邻的511个小区分开，这些I，Q序列有一个特定的偏移 这样确保了移动台不会解码其它基站的拥有相同Walsh code 的信道 I 与 Q 信号 与每个扇区的64个信道在模拟板中相互叠加产生每个扇区的复合“I”与复合“Q” 开销信道（道频信道、同步信道和寻呼信道）对于复合信号的贡献是固定的。前向业务信道对于复合信号的贡献与移动台到基站距离有关系 + Gain Control + Walsh code + Sync Channel + Gain Control Composite “I” + S Walsh code Composite “Q” + S Paging Channel(s) + Gain Control + Walsh code + Forward Traffic Channel(s) + Gain Control + 57

前向信道解调 IS-95A/J-STD-008 需要至少四个处理单元，他们可以独立的工作 其中三个单元必须能够解调多路信息 Correlator 1 Correlator 2 Correlator 3 Search Correlator De-Interleaver Decoder Vocoder Speech Output Combiner Mobile Receiver IS-95A/J-STD-008 需要至少四个处理单元，他们可以独立的工作 其中三个单元必须能够解调多路信息 另一个单元必须作搜索器，它用来搜索和估计每一个导频信道的PN 序列的偏移 58

 CdmaOne前向链路数据流方框图 I Q 复 用 59 W0 导频“1” W31 I信道导频伪码 同步信道 1200bit/s 卷积编码 交织 4800bit/s I Wi 寻呼信道 9.6kbit/s 4.8kbit/s 2.4kbit/s 19.2kbit/s 卷积编码 交织 Q 长码发生器 寻呼信道长码掩码 Wj 功率控制比特 复 用 Q信道导频伪码 业务信道 9.6kbit/s 4.8kbit/s 2.4kbit/s 1.2kbit/s 19.2kbit/s 卷积编码 交织 长码发生器 用户k的长码掩码 59

将Bits 转化为 Symbols Bits Symbols 在20 毫秒的业务帧内的比特包括下面的一项或多项 话音信息 (来自于声码器) 信令消息 第二部分业务信息 当把前向纠错算法用于这些信息比特时， 0s和1s被称为 symbols bit和symbol又复杂的多对多关系 1个bit 的信息被分配在多个symbol中 1个symbol 承载的信息有来自于多个bit 所有前向信道帧包含384个symbol 所有后向信道帧包含576个symbol Bits Symbols Forward Error Correction Creation of a CDMA signal begins with bits from the user’s vocoder. These bits are a compressed representation of the user’s voice activity during the last 20 millisecond “frame”. The job of the system is to deliver these bits undamaged to the other end of the CDMA link. The maximum bit rate is 14,400 bits/second (Rate Set II, 13 kB. vocoder). The first step in generating the CDMA signal is to protect the bits using a forward-error-correction strategy. Bits are fed through a convolutional encoder and a block interleaver. This process embeds the value of each bit into a larger number of symbols, making the signal much more robust and able to tolerate noisy corruption of the symbol stream without the loss of any of the original bits. The symbol rate emerging from this process is 19,200 symbols/second. 60

将Symbol分布到Chip中 Symbols Chips Symbol被变换成为特殊的64-chip模式用于传输 这里有64个这样的模式称作“Walsh codes” 在前向链路中，每个用户的symbol流（码信道）被分配使用这些模式中的一个 每个‘0’ symbol 被selected pattern （Walsh code ）替代 每个‘1’ symbol 被selected pattern的逻辑非替代 在反向链路中，所有的64模式（而不是他们的逻辑非）被用于每一个码信道 每6个symbol为一组，将其看作值域为[0，63]的二进制数，并用相应行的Walsh code替代 Symbols Chips Coding and Spreading The next step is to apply the CDMA spreading sequences. This raises the signal rate to 1,288,800 chips/second, thereby spreading the signal bandwidth and ensuring that there will be a spreading gain “payoff” when the signal is recovered at the other end of the link. Another very important consequence is that the spreading sequences embed the user’s unique identity into the chip stream, allowing multiple users to coexist and ensuring that only the correct receiver at the other end of the link will be able to extract this users information. The fast chip stream modulates the CDMA radio transmitter and the signal is radiated. While there are slight variations in the process on the forward and reverse links, this explanation accurately conveys the general principles for both links. 61

恢复过程 Bits Symbols Chips 在恢复过程中, 首先用下面的方式恢复symbol Despreading (integraton) Bits Viterbi Decoder 在恢复过程中, 首先用下面的方式恢复symbol 在 前向信道 内，移动台用给他分配的与接收信号作相关运算（集成64个的功率），一个与被分配Walsh code 完全匹配的接收信号对应‘0’ symbol ；一个与被分配Walsh code 完全不匹配的接收信号对应‘1’；介于两者之间的，移动台将根据相关后的结果更接近于那一方，作相应的判断。 在 反向信道 内， BTS寻找与接收信号最匹配的Walsh code的行号，并用这6位2进制数作为被发送的6个 symbol 当在20ms帧内的所有symbol被恢复时，Viterbi解码器被用于推测与 这个symbol块最可能对应的bit块（帧） 然后, 计算这一帧的CRC决定这个推测是否正确；如果不正确，这一帧将被抛弃（或“抹去”） 62

导频信道的产生 Walsh 函数0序列用于导频信道 使用short PN序列偏移产生512截然不同的 导频信道 Gain Control Baseband Filter I PN Q PN 1.2288 Mcps Walsh Function 0 Pilot Channel (All 0’s) Walsh 函数0序列用于导频信道 使用short PN序列偏移产生512截然不同的 导频信道 给的那些 pilot PN 序列的PN偏移索引值（包括0-511）乘以64就得到了实际偏移值 例如：15 (偏移索引) x 64 = 960 PN chips 结果：这个 pilot PN 的开始将移动 960 chips x 813.8 ns/chip = 781.25 µs 其它的前向或反向码信道都执行的“分布积分” 与“基带滤波” （未被表示），将在后面讨论 63

导频信道的捕获 移动台首先自己产生I和Q short PN 序列，然后在每一个可能的偏移位置用他们与接收到的混合信号作相关运算 00...01 导频信道 (Walsh Code 0) 移动台首先自己产生I和Q short PN 序列，然后在每一个可能的偏移位置用他们与接收到的混合信号作相关运算 在不长于15秒内 (通常是2到4秒) 所有的可能偏移 （32,768）被检查一次 移动台找出相关性最好的偏移（它的Ec/I0最好） 移动台锁住最好的导频（那个偏移时刻的Ec/I0最好），并且 辨认短序列开始的模式 （一个‘1’ 跟在15个连续的‘0’之后） 现在移动台已经准备好用Walsh code 32 作相关运算来捕获同步信道 64

同步信道的产生 每个同步信道帧中有32bit （1200 bps x 0.02666... 秒） Gain Control Baseband Filter I PN Q PN 1.2288 Mcps Walsh Function 32 1200 bps Block Interleaving R = 1/2, K = 9 Convolutional Encoding & Repetition 4800 bps bits modulation symbols chips 每个同步信道帧中有32bit （1200 bps x 0.02666... 秒） 速率1/2的卷积编码器使比特率加倍，因此0s与1s被称为“code symbols” 每个同步信道有64个code symbols 重复过程使速率再次加倍，每一个符号重复在此被称为一个 “modulation symbol” 每个同步信道有128个 modulation symbol Walsh code #32的4个拷贝扩频每一个 modulation symbol ，结果使速率增加a x256倍，现在0s和1s被称为“chips” 每个同步信道有32,768个chip 65

反向信道 66

CdmaOne反向业务信道数据流方框图 I Q 话音 编码器 卷积编码 符号 重复 交织 64元 正交调制 PNI 长码发生器 PNQ 67 9.6kbit/s 4.8kbit/s 2.4kbit/s 1.2kbit/s 28.8kbit/s 14.4kbit/s 7.2kbit/s 3.6kbit/s 28.8kbit/s 28.8kbit/s 307.2kcps 话音 编码器 卷积编码 符号 重复 交织 64元 正交调制 PNI I Q 1.2288Mcps 长码发生器 长码掩码 PNQ 标识用户身份 67

Access Channel Long Code Mask 接入信道的产生 28.8 ksps Convolutional Encoder & Repetition R = 1/3 1.2288 Mcps Access Channel Long Code Mask Long PN Code Generator Orthogonal Modulation 307.2 kcps Q PN (No Offset) I PN (No Offset) D 1/2 PN Chip Delay Block Interleaver Access Channel Information (88 bits/Frame) 4.8 kpbs Direct Sequence Spreading 消息试图任意排列以减少冲突的可能性 两种类型的消息 响应消息（响应基站消息） 请求消息（ 移动台自动发送） 68

速率1/3 卷积编码器 + g0 g1 g2 1 2 3 4 5 6 7 8 69 Code Symbols (OUTPUT) Information bits (INPUT) Code Symbols (OUTPUT) 1 2 3 4 5 6 7 8 69

接入信道的长码掩码 接入信道通过长码使其扩展，该长码叠加上掩码形成偏移。掩码结构如下： 1 这里 ACN ：接入信道号 PCN ：与其对应的寻呼信道号 BASE_ID ：基站识别号 PILOT_PN ：导频信道的short PN code偏移索引 1 PCN ACN BASE_ID PILOT_PN 41 33 32 28 27 25 24 9 8 70

(Ack response Timeout) Access Channel 探帧 Access Probe 1 Access Probe 1 + NUM_STEP (16 max) System Time TA RT PI IP (Initial Power) ACCESS PROBE SEQUENCE Select Access Channel (RA) initialize transmit power (Power Increment) (Ack response Timeout) (Probe Backoff) 71

反向业务信道的产生 72 I PN 1.2288 Mcps 9600 bps 4800 bps 2400 bps 1200 bps or 28.8 ksps R = 1/3 1.2288 Mcps User Address Mask Long PN Code Generator Orthogonal Modulation Data Burst Randomizer 307.2 kcps Q PN (no offset) I PN D 1/2 PN Chip Delay Direct Sequence Spreading R = 1/2 Convolutional Encoder & Repetition Block Interleaver This block flowchart shows the overall coding and signal flow involved in generating one user’s stream of chips transmitted on the traffic channel at the mobile. This entire process exists in the mobile. The similarly-generated signals of all other mobiles operating in this sector area all arrive at the BTS and are individually decoded by their respective assigned channel elements. 72

反向业务信道的 公共长码掩码 （基于 ESN） 反向业务信道使用公共长码掩码扩频。 它的结构如下： 1100011000 Permuted ESN-S 41 32 31 0 另外的选择是使用基于128位共享秘密数据寄存器的通用容器的“private long code mask” 73

Power Control Decision 反向信道的解调 BTS Receiver BSC Demodulator Search Correlator Demodulator Search Correlator Viterbi Decoder Speech Output Combiner De-Interleaver Vocoder Demodulator Search Correlator Power Control Decision U/D Command Demodulator Search Correlator IS-95A/J-STD-008要求一个移动台调制过程的补充过程 CDMA从多径成分获得的处理增益 从几个接收单元得到的接收信号可以和成为一个改善了接收质量的接收信号 74

CDMA中关键技术 话音激活技术 多径技术 功率控制(以后介绍) 软切换(以后介绍) 75

话音激活技术 原理：利用人说话的间歇性，在人话音停顿的 时候降低发射信号的功率 目的：降低多址干扰，提高系统容量 手段：可变速率话音编码 原理：利用人说话的间歇性，在人话音停顿的 时候降低发射信号的功率 目的：降低多址干扰，提高系统容量 手段：可变速率话音编码 76

无线信道中的多径衰落 77

S 手机中有什么？ Digital Rake Receiver Traffic Correlator Receiver RF Section IF, Detector Transmitter Vocoder Digital Rake Receiver Traffic Correlator PN xxx Walsh xx Pilot Searcher Walsh 0 Viterbi Decoder CPU Duplexer Digital Section Long Code Gen. Open Loop Transmit Gain Adjust Messages Audio Bit Packets Symbols Chips RF AGC S This simplified diagram of a handset shows the inner workings of both the transmitter and the receiver. Received BTS signals are passed into the receiver RF and IF sections. There the received chips are analyzed by a team of three traffic correlators called the “rake receiver”, and one pilot search correlator. Each traffic correlator (“rake finger”) can be set independently to any desired PN offset and any Walsh code. This allows the fingers to “surf” for individual multipath signal components from a single BTS sector, or in a soft or softer handoff, to look at the signals from different BTS sectors. At the end of each frame, each finger’s output is summed and processed by the Viterbi decoder, which feeds recovered vocoder bits to the vocoder and any signaling bits to the handset’s CPU. Meanwhile, the pilot searcher is constantly measuring active, candidate, and neighbor pilots, and even screening all remaining pilots for possibly useful signals which the handset might ask to use in soft handoff. Notice that all this digital magic is fed by a single-frequency radio receiver; the handset cannot notice other frequencies at the same time. The handset user’s speech is vocoded and digitally encoded before spreading and CDMA transmission. Notice the two independent mechanisms (open loop and closed loop transmit gain adjust) which control the gain of the transmitter chain. Open loop is driven by the receiver’s AGC voltage, while closed loop comes from the decoded BTS’ 800-per-second power commands. 78

Rake接收机 每一帧，手机使用三个业务信道的相关器的合成输出（（“rake fingers”） Handset Rake Receiver RF PN Walsh Searcher PN W=0 S Voice, Data, Messages Pilot Ec/Io BTS 每一帧，手机使用三个业务信道的相关器的合成输出（（“rake fingers”） 每个finger能够独立的恢复一个特定的PN偏置和Walsh code Fingers能被在延迟的多径反射，或者甚至是不同的BTS上捕获 Searcher不断的检查导频 It’s important to understand that the rake fingers actually extract traffic bitstreams from the channels they track, and therefore have full-time work to perform. At any given moment, the three fingers ideally “park” on the three most useful identifiable multipath components from among the one or several BTS sectors being demodulated. In contrast, the pilot searcher is a nomad, wandering from PN offset to PN offset making spot measurements of Ec/Io. Since there is no message bit stream on a pilot, the searcher never recovers any messages; instead, its whole job is just to provide quick reports of how good or how bad individual pilots appear. Incidentally, a BTS channel element also uses a “rake receiver” configured of four correlators which can look for multipath components from the handset. Handsets don’t transmit a pilot signal, so the BTS does not need a pilot searcher. Its four rake fingers are self-directed while they seek usable signal fragments from the handset. In a softer handoff, one channel element is sufficient to handle the generation and reception of the CDMA signal on all involved sectors. When a softer handoff is underway, the channel element’s four rake fingers are dedicated to whatever sectors have the most promising incoming signals from the handset. 79

多径接收  “Rake” 合路分路单元 信道译码器 0 1 2 3 3- 0 解 扩 3- 1 解 扩 3- 2 0 1 2 3 合路分路单元 3- 0 解 扩  3- 1 解 扩 信道译码器 3- 2 解 扩 解 扩 80

谢谢合作 81