Scrambling Codes
Once the different channels have been spread with appropriate channelization codes, they are combined, as shown in Figure 6-2, and then scrambled by a particular scrambling code. Two types of scrambling codes exist—long and short scrambling codes, with 224 possibilities for each type. The choice of a particular code is determined by the type of physical channel in question (as we shall see, other physical channels are available besides the DPDCH and DPCCH) and by the higher layer that requires the use of a channel in the first place. Depending on higher-layer requirements, a DPDCH or DPCCH can use either a short scrambling code or a long scrambling code. Clearly, the channelization codes are far from random, and they do not need to have pseudo-random properties. Scrambling codes, however, must appear to be random and thus must have pseudo-random characteristics. The easiest way to generate a pseudo-random sequence is through the use of a linear feedback shift register, such as that shown in Figure 6-4. This is basically a set of flip-flops that are clocked at a particular rate and the output of the last flip-flop is copied back into one or more of the other flip-flops, possibly after an addition. In Figure 6-4, each of the gain values (gn) is simply a 1 or a 0. Depending on the values of each gn (that is, exactly where the output is fed back), a different output pattern can be achieved. This output pattern can be described by a polynomial, known as a generator polynomial. It is possible to produce maximum-length sequences, known as msequences. This means that if a register has m elements, then it can produce a sequence of length 2m 1. For example, if a shift register has 10 elements, then it can produce a sequence of length 210 1 (1,023). This is a pattern that repeats after every 1023 bits. An m-sequence has a number of properties, including the property that, over the period of the sequence, there will be exactly 2m1 ones and 2m1 1 zeros. The long scrambling codes used in WCDMA are known as Gold codes and are constructed from the modulo 2 addition of portions of two binary m-sequences. The portions used are segments of length 38,400. This is due to the fact, as shall be explained later in this chapter, that the frame length in WCDMA is 10 ms, which corresponds to 38,400 chips. Because the long scrambling codes are generated from m-sequences, they have pseudorandom characteristics. The short scrambling codes also have pseudorandom characteristics. These, however, are much shorter, at a length of length 256 chips. Long scrambling codes are used in the case where the base station uses a rake receiver. Short scrambling codes can be used when the base station uses advanced multi-user detection techniques such as a Parallel Interference Cancellation (PIC) receiver. 232
1525 times read
|