汉服美女
Considerations
Author Department Phone Number
Chu Rui Chang Global Network Design
(Dept. 2779)(972) 685-7067 ESN 445-7067
2.1 CDMA Rever Link Capacity
The rever link pole capacity per ctor can be estimated by Eq.(2.1):
(2.1)
Where:
PG = processing gain;
百合花种子SG = Sectorization gain;
VAF = Voice Activity Factor;
韩国电视剧人鱼小姐
K = number of ctors per cell;
1/(1+f) = frequency reu factor;
Eb/No = Energy per bit to total noi spectrum density.
In the following, some key concepts will be explained.
(1) Orthogonality in the Rever Link
IS-95 rever links are not orthogonal. Maintaining orthogonality among channels requires channels from all handts to be synchronized at all time, when arriving at the BTS. This is difficult since one handt may be located within 50 meters from the BTS but another one may be 15 km from the BTS, thus they will have quite different propagation delays. Two channels with >= 1 chip pha difference will be unsynchronized and act like noi to each other. Therefore, becau of the pha uncertainties caud by differences in propagation delays, rever links are not orthogonal.
(2) Process Gain (PG)
Since the rever links are not orthogonal and all handts u the same frequency, every handt’s
Tx energy acts as co-channel interference to every other handt. Thus the SIR of any loaded CDMA system (before despreading) is almost always negative, which means the interference + thermal noi level is higher than signal.
It is the processing gain (PG) that makes CDMA systems work. Processing gain happens after the de-spreading process. It does not give any “gain” to the signal power, but does produce a “loss” to the interference power. The net effect is that there is a gain to the SIR.
Fig.3.1 shows the physical meaning of processing gain.
(A) Before the spreading, signal energy (marked black) and interference
energies (marked other colors) are superimpod on top of each other, indistinguishable. Interference power > signal power and SIR < 0.
Assume the bandwidth before spreading (bw) = 14.4 kHz.
(B) After spreading, both signal and interference are spreaded into
wideband (BW=1.25 MHz). Still they are superimpod on top of each other, indistinguishable. Interference power > signal power and SIR < 0.
(C) After de-spreading, only the signal is recovered into narrow band
(bw=14.4 kHz), all interference powers stay wideband (BW=1.25 MHz).
(D) After passing through a narrow band-pass filter (BPF, bw=14.4 kHz),
signal will pass through this BPF so its energy is unchanged. But most of the interference energy will be filtered out, only tho inside the
14.4 kHz will pass the BPF. Now the interference energy becomes
much lower, and we have signal power > interference power and SIR>0. The “loss” in interference power, is therefore BW/bw =
(1.25X10^6)/(14.4X10^3) = 19 dB. The processing gain for 14.4 kbts
channel is thus 19 dB.
SIR_after_despreading (Eb/No) = SIR_before_despreading * Processing_Gain
(A) (B)
(B) (D)
Fig. 2.1 The physical meaning of processing gain. (A) Before spreading; (B) After spreading; (C) After de-spreading; (D) After the band-pass-filter. (3) Eb/No
Eb/No is the energy per bit to total noi spectrum density ratio. This is the signal-to-(noi + interference)-ratio after the despread. It contains the processing gain:
寓学习于游戏
Rever link requires Eb/No to be about 7 dB to achieve the required performance. If the processing gain PG = 19 dB, then S/(I+N)before de-spreading can be as low as –12 dB.
(4) Sectorization Gain (SG)
This is the capacity gain produced from spatial filtering. If an omnicell could be divided into three equal ctors without any overlap, then each ctor would only receive 1/3 as much interference, thus the SIR, or capacity, would increa by 3 times. By the same argument, a 6-ctor would increa the capacity by 6 times, and so on.
In reality, ctorization will never be perfect, there will always overlap between ctors and thus interference from adjacent ctors. The narrower the ctor, the more difficult to control the interference from neighbor ctors. For a 3-ctor configuration, the ctorzation gain is about SF = 2.2 to 2.5. For a 6-ctor configuration, the gain is about SF = 3.5 to 4.5. The antenna 3dB beamwidth, the terrain and surrounding conditions all affect the ctorization gain that can be achieved in a real environment.
The value SG/K < 1 indicates that one ctor’s capacity will be smaller than that of an omni cell becau of the interference from adjacent ctors.
5月20日是什么日子(5) Voice Activity Factor (VAF)
When a subscriber talks, the channel transmits higher power, when subscriber paus, the channel transmits lower power (merely maintain the link). Thus on the average, a handt’s Tx power is lower than the peak power, and the interference to other urs is also lower.
Becau of the power control messages carried in the forward link, the VAF of forward link will be higher than VAF of the rever link. On the average, forward link VAF = 0.5 and rever link VAF = 0.45.
(6) Percentage of Out-of-cell interference to In-cell Interference (= f) Some times f is defined in another factor F=1/(1+f), the so-called frequency reu factor. f itlf is defined as
五百字Three assumptions are made to calculate the value of f:
(1) There are infinite number of cells in the system;
(2) All cells have equal size;
(3) All cells are equally loaded and subscriber’s distribution is uniform Total out-of-cell interference =
陈馀
Tot. interference contribution from the 1-st tier neighbor cells (6 cells)+ Tot. interference contribution from the 2-nd tier neighbor cells (12 cells) +
Tot. interference contribution from the 3-rd tier neighbor cells (18 cells) + …
Tot. interference contribution from the n-th tier neighbor cells + ……This is an infinite ries (Fig.2.2) and the sum of the ries gives the total out-of-cell interference.
Fig 2.2: Total out-of-cell interference is the sum of interference contribution from each tier of neighbors.
丑八怪薛之谦Note that the number of cells in k-th tier is proportional to the circumference of the k-th tier, which is proportional to D (= distance from the cluster center to the center of the k-th tier). If the path loss exponent is S, then each term in
the ries is proportional to
Ca 1: Path loss exponent S = 4: The ries converges the same speed as and the sum is about 65% as the in-cell interference. So f = 0.65, F =1/(1+f) = 0.6.
Ca 2: Path loss exponent S = 2: The ries diverges to infinite, at the same speed as . So the total out-of-cell interference is infinite and the CDMA capacity is zero. Although there are no real systems with infinite number of cells with path loss truly equal to 2, but this example does nd an important message. That is, if the path loss is too small, the capacity of the CDMA system will be very low. As was mentioned in Chapter I, if one designs BTS antennas too clo to each other without reducing the antenna height properly, the path loss will be very small and the system capacity degrades significantly.
(6) Pole Capacity:
Pole capacity is the theoretical maximum rever link capacity a CDMA system may achieve for a given Eb/No. Theoretically, when capacity approaches this limit, the receiver loading approaches 100% and the cell radius shrinks to zero. However, in real life systems, the number of active mobiles can increa while still maintain the call with a reduced Eb/No and incread FER, until the FER become so poor that all calls drop. Sprint operates at 50% of pole capacity (with Eb/No = 6.2 dB) as an average value, and 70% as the maximum.
2.2 CDMA Forward Link Capacity
Eq.(2.2a, b, c) gives three different forms of CDMA forward link capacity: (2.2a)
(2.2b)
(2.2c)
Where
N= the number of average urs a ctor can support assuming the above conditions.
