5.1.1Concepts of Power Control 概念

5Related technology

5.1power control

5.1.1Concepts of power control （概念）

Power[U1] control is a technique which allows the base station or the user to adjust the transmit power.

In[U2] the event that power control is not used, all mobiles transmit signals toward the base station with the same power, without taking into account the fading and the distance from the base station. Mobiles that are closer to the base station will cause significant interference to the mobiles that are further from the base station because of non-zero cross-correlation between signature sequences assigned to users. This effect is known as near/far effect.

Another[U3]reason for the use of power control is to prolong battery life by using a minimum of transmitter power to achieve the required transmission quality.

According[U4] to the facts mentioned above, for proper operation of a modern high-capacity cellular radio system, power control is an essential feature.

In[U5] order to realize power control, there are two important aspects to be considered:

1. Extracting relevant information from the available measurements.
2. Designing an appropriate power control algorithm and tuning the optimized parameters.

When[U6] considering the power control in real systems, the following aspects are interesting:

1. Quality[U7] measure: Speech quality is a very subjective quantity.Signal-to-Interference Ratio (SIR) is extensively used as an adequate objective measure in previous works, even though it is far from ideal. Because in the case of data transmission, bit error rate (BER) requirements can be very stringent and SIR might not be an adequate measure. If the signal and interference powers are constant，BER and SIR contain equivalent information regarding quality. But in real systems the SIR is time variant and thus the average SIR will not correspond to the average BER. In this case BER is a better quality measure.
2. Available[U8] measurements: Usually the measurements are given in reports comprising a quality indicator (QI) reflecting the quality and a received signal strength indicator (RSSI) reflecting the received signal strength at the receiver. These values are coarsely quantized in order to use few bits.
3. Constraints[U9]: The output power levels are limited to a given set of values due to hardware constraints. This includes quantizing and the fact that the output power has an upper and a lower limit.
4. Time [U10]delays: Measuring and control signaling take time, which results in time delays in the network.

Also[U11], in real systems, a number of undesired effects need to be heeded:

1. Measuring and controlling signals take time, which results in time delays in the system.
2. The possible output powers of the transmitters are constrained due to physical limits and quantization.
3. The signals needed for power control may not be available and have to be estimated.
4. Quality is a subjective measure and relevant objective measures have to be employed.

5.1.2Classification of power control （分类）

Power control techniques can be classified in many different ways. It is shown as follows.

.

Figure 6-1 Classification of power control

5.1.3Power control in TD-SCDMA

5.1.3.1General description

Table 6-1: TPC [U12]characteristics

Uplink / Downlink
Power control rate
（速率） / Variable
Closed loop: 0-200 cycles/sec.
Open loop: about 200us – 3575us delay / Variable
closed loop: 0-200 cycles/sec.
Step size（步长） / 1,2,3 dB (closed loop) / 1,2,3 dB (closed loop)
Remarks（备注） / All figures are without TPC decoding and received power measurements. / Within one timeslot the powers of all active codes may be
balanced to within a range of [20] dB

5.1.3.2Principle of power control （原理）

5.1.3.2.1Uplink Control

5.1.3.2.1.1Open loop power control for the UpPTS（上行开环）

The transmit power level by a UE on the UpPTS shall be calculated based on the following equation[4]:

开环发射功率＝信道损失+期望的接受功率

PUpPTS = LP-CCPCH + PRXUpPTS,des

Where, PUpPTS : transmit power level in dBm,

LP-CCPCH: measured path loss in dB,

PRXUpPTS,des: the desired RX power level at cell’s receiver in dBm.

5.1.3.2.1.2Common Physical Channel

The transmitter power of UE[U13] on P-RACH [U14]shall be calculated by the following equation [4]:

PPRACH = LP-CCPCH[U15]+ PRXP-RACH, des

where,PPRACH : transmitter power level on the P-RACH ;

PRXP-RACH, des: the desired receive power level on the P-RACH.

5.1.3.2.1.3Dedicated Physical Channel

The[U16] closed loop power control makes use of layer 1 symbol in the DPCH（专用物理信道）[4]. The power control step can take the values 1,2,3 dB within the overall dynamic range 80dB. The initial transmission power of the uplink DPCH is the same power level used in last P-RACH transmission.

Closed-loop [U17]TPC is based on SIR, and the TPC processing procedures aredescribed as follows. During this power control process, the nodeB periodically makes a comparison between the received SIR measurement value and the target SIR value. When the measured value is higher than the target SIR value, TPC command=’down’. When this is lower than the target SIR value, TPC command = ‘up’. At the UE, soft decision on the TPC bits is performed, and when it is judged as ‘down’, the mobile transmit power shall be reduced by one power control step, whereasif it is judged as ‘up’, the mobile transmit power shall be raised by one power control step. A[U18] higher layer outer loop adjusts the target SIR.

5.1.3.2.2Downlink Control

5.1.3.2.2.1Common Physical Channel

The power of the P-CCPCH[U19]:

The primary CCPCH transmit power is set by high layer signaling and can be changed based on network determination. The reference power of P-CCPCH is signaled on the BCH on a periodic basis.

The power of the F-PACH: 寻呼信道

It is set by the network.

The power of the S-CCPCH: 公共控制物理信道

The power of the FACH（前向接入信道） in S-CCPCH:

It is set by the network or can take into account both the received power level on the P-RACH（随机接入信道） from the addressed UE and the transmit power level as signaled by the UE.

The power of the PCH in S-CCPCH

This condition is the same as P-CCPCH

5.1.3.2.2.2Dedicated Physical Channel

The[U20] initial transmission power of the downlink Dedicated Physical Channel is the same as the FACH in S-CCPCH or set by the network until the first UL DPCH arrives. After the initial transmission, the node B transits into SIR-based closed-loop TPC.

The measurement of received SIR shall be carried out periodically at the UE. When the measured value is higher than the target SIR value, TPC command =’down’. When this is lower than the target SIR value, TPC command = ‘up’. At the Node B, soft decision on the TPC bits is performed, and when it is judged as ‘down’, the transmission power shall be reduced by one step, whereas if judged as ‘up’, the transmission power shall be raised by one step.

5.1.4Algorithms of power control 算法

5.1.4.1Basic method of power control

The basic method of power control is as follows:

First, measure the received SIR.

Secondly, compare the received SIR with the target SIR value.

Then, generate power control bits.

Next, get power control bits from the received data.

Finally, adjust transmit power.

5.1.4.2Linear prediction-based closed-loop power control

The target of this algorithm is to mitigate the impact of the errors in TPC bits and make up for the delay of TPC bits [13]. The whole process is described in Figure 6-2.

When the mobile station gets the power control bits from the base station, it doesn’t use them to adjust the transmit power immediately. It will first accumulate the power control bits. Then put the accumulation into a linear predictor. Linear predictor will predict the accumulated power control bits.

Then subtract the previous prediction value from the current one. The result will go through a multi-level PWC( piecewise constant ). The output of PWC is the value for adjustment.

And multi-level PWC is defined as follows:

Here, d(n) is the adjustment of transmit power,

s(n) is the current prediction value,

s(n-1) is the previous prediction value,

is power control step, T1,T2,T3 can be set by system.

Figure 6-2 The Multi-level PWC with linear predictor

5.1.4.3Strength and SIR-combined power control

The target of this method is to alleviate positive feedback in SIR-based power control [14]. The principle is shown in Figure 6-3.

Figure 6-3 The principle of strength and SIR-combined power control

There are four circumstances. And different methods are adopted.

When SIR and strength are both below threshold (This situation is denoted as (0,0)), that means the performance is not satisfactory and the strength of the signal is still very low. So increase the transmit power.

When SIR and strength are both above threshold (This situation is denoted as (1,1)), that means the performance is satisfactory and the strength of the signal is high. So decrease the transmit power.

When SIR is satisfactory and strength is below threshold (This situation is denoted as (1,0)), take no action because SIR reflects QoS and the transmit power needn’t to be changed.

When strength is satisfactory and SIR is below threshold (This situation is denoted as (0,1)), that means performance of system is not good, we should be cautious because at this time the strength of the signal already exceeds its threshold, it is very likely to cause positive feedback. So take special action.

Specific steps are shown as follows:

First, record the number of consecutive (0,1) before current (0,1) and denote it as w.

Then, postulate a rule. For example,

Here, r is a constant parameter decided by the system,

is the original power control step,

is the adjusted power control step.

So according to the formula above, we can mitigate the positive feedback of SIR-based power control.

5.1.4.4Utility-based power control

The target of this algorithm is also to alleviate positive feedback in SIR-based power control [16]. And the method is using the utility function for every user.

Denote Ui as the utility function of user i, Ci as the cost function of user i and NUi as the net utility of user i. The relationship between them is NUi=Ui-Ci. Our goal is to maximize NUi by adjusting the transmit power.

5.1.4.5Integrated power control and base station assignment

The principle of this algorithm is integration of power control and base station assignment [15].

Both the transmit power and base station assignment should be updated.

But this method has some problems, such as high communications overhead in determining the minimum transmit power and selecting base station, large storage capacity of BS and high computational requirement of MS

5.1.5SIR measurement methods

5.1.5.1General methods for SIR measurement

5.1.5.1.1Mean value and variance method

In CDMA system, the sum of the MAI produced by many users is well approximated as complex Gaussian noise according to the central limit theorem. This means that the MAI can be combined with background noise characterized by additive white Gaussian noise (AWGN) and can be treated as composite Gaussian noise.

Suppose Y=S+I

The average power of signal is.

If S and I is uncorrelated, then and .

So and

We have supposed the mean value of I is zero, so and

And the estimation process is as follows:

Figure 6-4 The process of SIR estimation based on mean value and variance

5.1.5.1.2Eigenvalue Decomposition method

5.1.5.1.2.1The principle of ED method

Assume a known L bits training sequence of target user [18]:

The symbol sequences transmitted by the other K users are:

The model of received signal:

Here, rj0 is the crosscorrelation between user j’s address code and user 0’s.

hij is the impulse response of a time varying multi-pathchannel from user i to receiver j.

Assume the covariance matrix of the received signal is:

(user 0 is target user)

Here,,

And is small and can be negligible. So

So

We denote the interference plus noise power by

If we denote and, then

Here is small and can be negligible, soand .

5.1.5.1.2.2Steps of ED method

First, make an eigenvector decomposition of the sample covariance matrix

where , with and .

Secondly, estimate the dimension of the signal subspace.

Then, estimate the interference power according to and t

Next, the signal power according to

T

Finally, the estimation of the SIR is then obtained as

where the factor 1/L accounts for having L sample of the received signal within observation vector.

5.1.5.1.2.3Estimation of signal subspace dimension

We use MDL(Minimum Descriptive Length) method [18].

Let

Define the sphericity test function as

and the MDL objective function, as

The signal subspace dimension is then estimated as the argument of the minimization problem. That is to say

5.1.5.1.3PASTd method

5.1.5.1.3.1The principle of PASTd method

Let x be a complex valued random vector process with correlation matrix R [18].

Consider the following scalar function

We can prove that J(W) has a global minimum at which the column space of W equals the signal subspace.At each stationary point of J(W), it equals the sum of the eigenvalues of those eigenvalueswhich are not involved in the signal subspace. All sample vectors available in the time interval are involved in estimating the

signal subspace at the time instant n.

If y(n) is the data input at the nth instant ,the cost function to be minimized is

where is a forgetting factor meant to discount the past observations. Iterativealgorithms are necessary to minimize J(W).

5.1.5.1.3.2Steps of PASTd method

First, choose and .

Here, is the initial eigenvalue.

Then, when t = 1,2…, y(t) = current data vector and .

Finally, when j = 1,…,d

Here, d is the signal subspace dimensionand it is assumed to be known.

are the columns of the matrix .

A simple initial choice for these columns would be the d leading unit vectors of the L×L identity matrix. ( L is the dimension of the input data vector y).

PASTd algorithm allows for explicitcomputation of the eigencomponents. is an estimation of the jth eigenvectorof the covariance matrix C(t) andis an estimation ofthe correspondingeigenvalue attime t. Thus, .

So the signal power is

The noise plus interference is

SIR is

5.1.5.2SIR estimation for TD-SCDMA

There are several methods to estimate SIR for TD-SCDMA. Since in TD-SCDMA we adopt joint detector instead of Rake receiver, we should use special methods based on joint detector.

Here, we can use two methods to estimate SIR.

5.1.5.2.1The first method for SIR estimation

Since the signal power from other users is regarded as interference, SIR can be defined as follows [19][23][24]:

For uplink

Here, S is the signal power of the target user; Gp is process gain; is the interference from the users belonging to the same base station as the target user; is the interference from the users belonging to the different base station from the target user; is the interference cancellation factor of joint detection; No is noise.

For downlink

Here, S is the signal power of the target user; Gp is process gain; is the interference from other downlink channels of the same base station; is the interference from other base stations; is the orthonormal factor; No is noise.

In [23][24], we can know that if joint detection is not used, can approximately equal to the sum of other users’ received power.

Before computing SIR, we should estimate and first. They can be got from experiments.

After that, we will make estimation following the below steps:

First, get CIR for each user from received midamble.

Secondly, compute signal power of each user from the corresponding CIR.

Then estimate noise and interference.

Finally, use the above formula to get SIR.

The block for SIR measurement is as follows.

Figure 6-5 The block for SIR measurement in TD-SCDMA

For estimating the signal power, there are several methods.

Method 1. [20]

k=1,…,K

Here,is the Walsh code for user k; is the channel estimation of antenna ka for user k; Ka is the total number of antennas; K is the total number of users; M is decided by the modulation method. is bit energy of the received data.

Then we can estimate signal power from .

That is to say

Here, T is the duration of one bit. is the signal power of user k.

Method 2 [21]

Here, h(k) is the channel estimation for user k; is the signal power of user k; denotes the Hermitian transpose of x..

5.1.5.2.2The second method for SIR estimation

We use ZF-BLE algorithm in joint detection [20][22]. Assuming that the CIRs of the links between transmitter and receiver are perfectly known at the receiver, then by applying the zero forcing block linear equalizer(ZF-BLE) a linear estimate

So SIR can be represented as follows [22].

Where the expectation is with respect to the data bits of MAI’s and noise. In the simulation, the expectation operation is replaced by the time averaging operation.

The block for SIR measurement is as follows.

Figure 6-6 The block for SIR measurement in TD-SCDMA

5.1.5.3Adaptive SIR-based power control

Power control step can be fixed or unfixed. We can change step in the simulation according to the performance. If the performance is too bad, we should improve transmit power by a larger step. If the performance is below desired but not very bad, we can just improve transmit power by a small step. Step available is 1dB,2dB and 3dB. We can compare SIR with target SIR, and then decide which step should be used.

5.1.6Simulation of power control

5.1.6.1Simulation of BER-based power control

BER-based power control is described in the following block.

Figure 6-7 The block for BER-based power control

5.1.6.2Simulation of SIR-based power control

We simulate SIR-based power control according to the following block.