Academic Open Internet Journal

Volume 15, 2005

DYNAMIC CLASS BASED POWER CONTROL FOR MEDIA STREAMING IN CDMA

Mariappan.V

(vbrms@yahoo.com)

Narayanasamy.P

(sam@annauniv.edu)

Department of Computer Science & Engineering,

Anna University, INDIA – 600 025.

Abstract

The objective of this paper is to establish the relationship between Quality of Service (QoS) requirements and the transmission power using dynamic class based power control method. In CDMA system, multiple users share the same bandwidth and they are separated by pseudo random codes; consequently each user is interfered with each other. Hence the need for power control is self-evident. This paper explores the relationship between QoS requirements and the transmission power using dynamic class based power control method. This method is implemented in Matlab using simulink and CDMA reference block set. The result shows that there is a significant performance improvement in terms of data users, coverage capacity and outage probabilities compared to the existing power control mechanism.

Key Words: 3G, CDMA, Power Control, Outage Probability.

1. Introduction

3G Wireless Networks has wide range of services. Data applications have fundamentally different Quality of Service (QoS) requirements and traffic characteristics than video or voice applications. CDMA has been recognized as a viable alternative to both Frequency Division Multiple Access (FDMA) and Time Division Multiple Access (TDMA). CDMA has many advantages such as Universal one-cell frequency reuse, Narrowband interference rejection, Inherent multipath diversity, Soft hand off capability and Soft capacity limit. But these advantages can be hindered by the increasing interference caused by other users, since all signals in CDMA system are sharing the same bandwidth and overlapping in time. It is essential to exercise some kind of control to maintain acceptable Signal – to – Interference Ratio (SIR) for all users. Hence maximizing the system capacity by minimizing the outage probability, which is the probability that a call will be dropped due to inadequate SIR level.

One critical problem with CDMA is the near far problem. This problem occurs in the absence of power control. If all mobiles are transmitting at the same power level, the mobile closest to the base station will overpower all others. Yet, another reason for power control is the battery lifetime. If the mobile station continuously transmits at a power higher than that to maintain an acceptable SIR, the battery lifetime will be reduced. Using power control, each mobile station can transmit at the minimum power needed for maintaining the required SIR ratio, thus conserving its battery life.

Transmission power control not only solves the near far problem, it also prolongs the mobile battery life and can be used to allocate different channel qualities to users having different QoS requirements. Several power control techniques have been studied. An eigen value problem for no-negative matrices [1] –[4], Iterative Power Control Algorithms [5][6], Optimization based approach [7]-[9] and other variations [10] –[12]. A useful framework for uplink cellular power control [13]. This paper addresses the problem of how QoS requirements are improved with minimal total transmission power using Dynamic Class Based approach.

The remaining part of this paper is organized as follows. Section 2 introduces the system model and its advantages. Implementation model is described in Section 3. Performance evaluation and results are presented in Section 4. Finally Section 5 constitutes conclusion and the future work.

2. Class Based Approach

CDMA systems have to support multimedia services like voice, data and fax. Issues in providing multimedia services on wireless include multiple access, bandwidth rationing and power control. Different services have different QoS requirements, power and rate constraints. In order to achieve the required QoS they can alter their power and/or rate constraints. Since users are interfered with each other, achieving each user’s QoS requirement is coupled with their power. This is formulated as a constrained optimization problem. Thus the users are split into classes based on their rate requirements and their transmitted power, which are calculated by the base station to achieve better performance of the entire CDMA cell.

2.1 Need For Class Based Power Control

Variety of services

The rate, QoS requirements and power needed for each user is different due to different types of services provided by CDMA such as digital picture, e-mail, fax, voice, GPS, etc. So fixed requirements for each user is no longer an efficient one. Classifying the users and providing the resources based on their needs is a must.

Interference

Each user operates with a high power in the system and hence there is interference inside the system. Some times it produces the problem such as near-far effect, dropping users, and increase in noise. In order to reduce interference inside the system, dynamic class based power control is mandatory.

Increase in capacity

Only a limited numbers of users can be used in a CDMA cell. By introducing class based power control in CDMA system the overall interference of the system reduces and there will be a room for more number of users to operate in a cell.

2.2 Dynamic Class Based Power Control Algorithm

Let N be the number of users in a CDMA system. Each user has QoS, power and rate constraints. The chip rate and the total bandwidth available (W) are fixed. Define the existing parameters in a CDMA cell as

Power vector P = (P₁, P₂, … P_n)

Rate vector R = (R₁, R₂, … R_n)

QoS vector V = (V₁, V₂, …V_n)

Each user specifies a minimum tolerable QoS. Usually this is either in terms of BER or FER. Here it is assumed that the BER / FER requirement can be mapped to E_b /N_o requirement [14]. User also specifies the maximum power limit that they can afford and minimum data transfer rate that they require. The required limits are

Power vector p = (p₁, p₂, … p_n)

Rate vector r = (r₁, r₂, … r_n)

QoS vector v = (v₁, v₂, …v_n)

The problem of finding the optimum power and rate vector can be summarized as follows:

Where

E_b = Bit error rate (BER)

N_o = Frame error rate (FER)

W = Total Bandwidth

h = Channel Gain

h₀ = Spectral density of Gaussian noise

In the above QoS equation the required QoS is achieved by two approaches. Keeping ‘R’ as constant value and trying to minimize P. Keeping ‘P’ as constant value and trying to maximize R. Optimize the QoS by using these two constraints. Thus this could be a constrained optimization problem.

Minimizing Total Transmitted Power

For a system with transmitted power vector P, the problem reduces to:

When the optimum rate vector is R*=(r₁, r₂, …r_n). The power vector can be obtained by solving the equation 3.

Get the required rate, QoS and power from each user and classify them according to their requirements. This classification is done based on their rate, which in turn depends on their service. Then based on their requirements calculate the optimum power in which transmitter should operate as per the users class and transmit it to the user.

2.3 Advantages Of Class-Based Power Control

Heterogeneity of users considered

In class based power control, different types of classes for different types of users are considered. In contrast to earlier method, which assumes fixed requirements for all users, this method uses different classification based on user needs.

Utilization of unused resources

In class based power control, the resources unused by the low-end users are distributed to high end users. Thus it supports the multimedia users to use the unused resources taken by the low end users.

Increased number of users

Power control is achieved by implementing the class based power control method, hence the overall interference and noise in the system gets reduced. It gives room for some more number of users to enter the system. The simulation result shows that there is a significant performance improvement in terms of number of users.

Better BER and FER for multimedia applications

Class based power control aims at optimizing the power but it never compromises in QoS parameter. So by optimizing power for a user reduces the Bit Error Rate and the Frame Error Rate.

3. Implementation

This dynamic class based power control method is implemented using CDMA Reference Blockset. The CDMA Reference Blockset is a collection of Simulink blocks designed to develop and simulate CDMA wireless communication system. The forward and reverse traffic channel model was constructed using the CDMA Reference Blockset.

3.1 Forward Traffic Channel - Base Station Transmitter Model

The basic components of this model are the transmitter, receiver, and the channel. The resulting raw (without channel coding) bit error rate (BER) is displayed in the simulation as a measure of performance. This model essentially implements:

Power bits from the class based power control algorithm are inserted in the power control commands of the Forward Traffic channel.
Spreading and Despreading of the data symbols are performed in the blockset.

The functioning of the transmitter section is to generate the modulated waveform that contains the various forward link channel components including the Pilot, Sync, Paging and other traffic channels. The transmitter components are the data source, the scrambler, the spreading and modulation, and the transmit filter.

The Data Source subsystem produces many data elements in this model. It consist of:

Random number generators that generate the bipolar random data symbols for the Traffic, Sync and Paging channels.
A signal that represents the data rate.
The Long Code Generator library block, which generates the long code used to scramble the data.
The Walsh Code Generator and Short Code Generator library blocks which generate the Walsh and complex PN codes. These codes are used for orthogonal spreading of the modulation symbol data. The Forward Channel Scrambler library block uses the decimated long code input to scramble the input traffic frame and insert power bits. The scrambled traffic channel data and the data symbols from the various other types of channel sources are input to the Spreading and Modulation subsystem. These inputs are orthogonally encoded by their respective Walsh codes, added, and spread with the in-phase and quadrature components of the short PN sequence. The Forward Channel Base Station Transmitter Interface library block accomplishes part of this, which is inside the Spreading and Modulation subsystem. The generated signal is processed by the pulse shaping Transmit Filter block, which generates the modulated I and Q waveforms. The Rayleigh Multipath and AWGN Channel subsystem simulates the propagation through multiple paths of a Rayleigh fading channel. Complex white Gaussian noise is added to the channel output. This noise represents the interference generated by other base stations that are using the same frequency band [16].

The receiver section is responsible for the recovery of the data symbols transmitted on the traffic channel. The operation performed in this section includes filtering, rake correlation, rake demodulation and descrambling. The Forward Channel Detector library block is a masked subsystem with several components inside. The rake receiver computes symbol duration correlations for the Traffic data and Pilot symbols. These correlation values are used by the Forward Channel Rake Demodulator to recover the Traffic channel symbols. The Traffic symbols are further processed by the Forward Channel Descrambler to obtain the decision values for the original transmitted data symbols.

This simulation uses the raw BER as the measure of the performance under the channel and noise conditions selected.

3.2 Reverse Access Channel - Mobile Station Transmitter Model

The Reverse Access Channel implements the calculation of the different power commands for the user in the CDMA cell with respect to other users of the cell and writes the modified power control bits to be used by the forward traffic channel. The basic components of this model are the transmitter, receiver, and the channel. This model essentially implements:

Sending of power, rate and QoS limits of the user to the Base Station from the Mobile Station through the Access Channel.
Spreading and despreading of the data symbols are performed with in the blockset.

The base station transmitter section performs the Cyclic Redundancy Check CRC) generation, convolutional encoding, symbol repetition, and interleaving. The Random Binary Frame Generator masked subsystem generates random data that acts as information bits. The Mobile Station Transmitter Data subsystem provides the selection of the data rate. The CRC Generator library block appends the CRC bits to the information bits. These CRC bits are used to detect errors in the data frame at the receiver. The Reverse Channel Convolutional Encoder library block convolutionally encodes the data using a 1/2-rate encoder for protection against channel errors.

While CDMA supports variable data rate operation, the data frame at this stage can have a number of different sizes. Depending on the data rate, the Reverse Channel Repeater/Derepeater library block may repeat the bits. It receives to create a data frame of 384 symbols. This interleaved data is converted to the bipolar form, and the AWGN Channel block adds white Gaussian noise to each symbol of the bipolar data. This noise represents the random error in the demodulation of the symbol. The mobile side performs the reverse operations.

The final metrics from the Reverse Channel Viterbi Decoder block are also input to the Frame Quality Detector block, which decides whether the frame was correctly received. The Frame Quality Detector block outputs the Quality Indicator signal, as well as the information bits without the CRC bits. One Error Rate Calculation block compares the information bits to the bits generated at the source, while another Error Rate Calculation block compares the Quality Indicator bits with zero. Finally, the resultant bit and frame error rates are displayed.

Once the power, rate and QoS limits are gained from the frame, they are used to calculate the target power for the user in the next frame. The algorithm for class based power control is implemented in C. The C file accesses the MATLAB default workspace whereas the details of all the users’ power, QoS and rate limits are stored in a table. The C file makes the computation and the resultant value is transmitted back to the MEX file that called it. This value is stored again in the MATLAB workspace and is used by the Forward Traffic Channel for setting the power control bits in the Scrambler subsystem.

The following parameters determine the function of the models described above and they can be modified to simulate the different operational environments.

Table 1: Parameters of Simulink Models

Data Rate	Full, Half, Quarter, and One-Eighth.
Rate Set	Rate Set I Rate Set II
Doppler Frequency in Channel	Doppler frequencies of the cellular mobile environment are in the 0 to 200 Hz range.
Signal-to-Noise Ratio. E_b/N_o	The detection improves with an increase in the signal-to-noise ratio.
Random Seed	To run the simulation with different seeds. Random Data Frame Generator, Random Power Bits, Multipath Rayleigh Fading, and AWGN Channel.
Channel Paths and Rake Fingers	Changes in the number or delay of fading channel paths and rake fingers must correspond to each other.

4. Performance Evaluation and Results

Here the parameters used for performance evaluation are total transmitted power and the number of users in the system. Figure 1 compares the total transmitted power in each method for a particular number of users. The number of user parameters in this algorithm is varied and the change in the total transmitted power is observed. The result shown in Figure 1 reveals that the increase in number of users in class based power control requires only a little increase in the total transmitted power as compared with the existing method. Thus the power needed is considerably reduced in class based power control method.

Figure 1 Number of users Vs. Total Transmitted Power

For various total transmitted powers the maximum number of users accommodated in the system are plotted in Figure 2.

Figure 2 Total Transmitted Power Vs. Target number of users

Figure 2 shows that there is a slight increase in total transmitted power which allows a great number of users to enter the system in class based power control method as compared against the existing method. This is because, class based power control method gets the unused resource from low end users and it can be used by high end users or some new users.

5. Conclusions and Future Work

This paper compares the performance of the existing system and the class based power control systems. Two metrics are used, based on which the comparison of systems are made. They are the total transmitted power and the number of users accommodated within the cell. This set of metrics should be measured for comparing sets of different CDMA cells of users. It was found that the class based power control algorithm yielded better results over the existing system in terms of both the metrics. Thus yielding the performance requirements and limits of the users resulted in better service to all the users of the cell. Additive white Gaussian noise was added to both the models constructed based on the number of users existing at that point in the cell. It could be further modified in the future to accurately reflect the existing noise in a CDMA cell, if profiles of such noise metrics are available. Thus the performance of the power control algorithm in real time situations could be accurately modeled. The effect of this power control algorithm on the reverse access channel needs to be explored. The other feedback characteristics from the mobile users could be exploited for use by the base station for performance enhancement.

References

[1] J.M.Aein (1973), ”Power Balancing in System Employing Frequency Reuse”, COMSAT Technical Review, vol. 3, no. 2, pp. 277 - 300.

[2] R.W. Nettleton and H.Alavi (1983), “Power Control for a Spread Spectrum Radio System”, in IEEE Vehicular Technology Conference, pp. 242 – 246.

[3] J.Zander (1992), “Performance of Optimum Transmitter Power Control in Cellular Radio Systems”, IEEE Vehicular Technology, vol. 41, pp. 57 – 62.

[4] S.A.Grandhi, R.Vijayan, D.J.Goodman and Zander (1993), “Centralized Power Control in Cellular Radio Systems”, IEEE Transactions of Vehicular Technology, vol. 42, pp. 466 – 468.

[5] G.J.Foschini and Z.Miljanic (1993), “A Simple Distributed Autonomous Power Control Algorithm and its Convergence”, IEEE Transactions of Vehicular Technology, pp. 541 – 546.

[6] S.V.Hanly (1995), “An Algorithm of Combined Cell-site Selection and Power Control to maximize Cellular Spread Spectrum Capacity”, IEEE Journal on Selected Areas in Communications, vol.13, pp. 1332 – 1340.

[7] D.Julian, M.Chiang and D.O.Neill (2001), “Robust and QoS Constrained Optimization of Power Control in Wireless Cellular Networks”, in Vehicular Technology Conference, Atlantic City, NJ, USA, vol.3, pp.1932 – 1936.

[8] S.Kandukuri and S.Boyd (2002), “Optimal Power Control in Interference Limited Fading Wireless Channels with Outage-Probability Specifications”, IEEE Transactions on Wireless Communications, vol.1, no.1, pp. 46-55.

[9] N.Bui and S.Dey (2002), “Optimal Power Control in CDMA Over Markov Fading Channels”, in Proc. IEEE International Symposium on Information Theory, (Lausanne, Switzerland), pp.79.

[10] R.D.Yates and C.Y. Huang (1995), “Integrated Power Control and Base Station Assignment”, IEEE Transactions of Vehicular Technology, vol. 44, no.3, p. 638 – 644.

[11] S.Grandhi, J.Zander and R.Yates (1995), “Constrained Power Control”, International journal of Wireless Personal Communications, vol. 1, no. 4.

[12] S.Ulukus and R.D.Yates (1998), “Stochastic Power Control for Cellular Radio Sysems”, IEEE Transactions on Communications, vol. 46, pp. 784 – 798.

[13] R.D.Dates (1995), “A Framework for Uplink Power Control in Cellular Radio Systems”, IEEE Journal of Selected Areas in Communications, vol.13, no.7, pp. 1341 – 1347.

[14] Ashwin Sampath, P. Sarath Kumar, Jack M. Holtzman (1995), “Power Control and Resource Management for a Multimedia CDMA Wireless System”, Proceedings of PIMRC pp. 21-25.

[15] Vijay K. Garg (2000), “IS-95 CDMA and cdma2000 Cellular/PCS Systems Implementation”, Prentice Hall PTR.

[16] Loutfi Nuaymi, Xavier Lagrange, Philippe Godlewski (2002), “A Power Control Algorithm for 3G WCDMA System”, European Wireless.

[17] Tim Holliday, Andrea Goldsmith, Peter Glynn (2002), “Optimal Power Control for CDMA Systems in the Wideband Limit”.

[18] Rintamaki, Virtej, Koivo (2001), “Two Mode Fast Power Control for WCDMA systems”, VTC IEEE.

[19] W. Song, B. Ahn, B. Kim, W.Kim, S. Kim, M.Choi (2002), “Evolutionary Computation and Power Control for Radio Resource Management in CDMA Cellular Radio Networks”, The 13th IEEE International Symposium on PIMRC.

[20] www.cdg.org/technology/index.asp

[21] J. Andrews, A. Agrawal, T. Meng, and J. Cioffi, “A Simple Iterative Power Control Scheme for Successive Interference Cancellation”.

[22] T. Alpcan, T.Basar, R. Srikant and E.Altman (2002), “CDMA Uplink Power Control as a NonCooperative Game”, Spring Lab Seminar.

[23] A. F. Almutairi, S. L. Miller, H. A. Latchman (2002), “Power Control Algorithm for MMSE Receiver Based CDMA Systems”, IEEE Communication Letters, Vol. 4, No. 11, pp. 346-348.

[24] CDMA Reference Blockset For use with Simulink, Users Guide, Algorex Inc., The Math Works

Technical College - Bourgas,