wireless packet transmission with variable processing gain control in packet-switched and...

11
Wireless Packet Transmission with Variable Processing Gain Control in Packet-Switched and Circuit-Switched Mode Integrated DS-CDMA Systems Takumi Ito, Seiichi Sampei, and Norihiko Morinaga Graduate School of Engineering, Osaka University, Suita, 565-0871 Japan SUMMARY A DS-CDMA system with both a circuit-switching scheme for voice transmission and a packet-switching scheme for data transmission is effective in multimedia communication required to transmit a variety of media. Although the Slotted-ALOHA system is believed to be an effective packet-switched system from the perspectives of transmission efficiency and interference power, only open loop transmit power control is used because of the difficulty in using closed loop transmit power control in the ALOHA system. In this paper, we analyze the performance of open loop transmit power control that uses a power margin to obtain the quality required even in packet-switched termi - nals, and propose wireless packet transmission with vari- able processing gain control as the method to achieve high-speed transmission by more effectively using the free channels. The results of computer simulations showed that the system capacity of circuit-switched terminals is not damaged when an appropriate power margin is set, and the additional effect of the packet-switched terminals is in- creased. In addition, the proposed system can greatly short - en the average delay time of packet-switched terminals. ' 2001 Scripta Technica, Electron Comm Jpn Pt 1, 84(10): 1626, 2001 Key words: DS-CDMA; packet switching; multi- media communication; open loop transmit power control; power margin; variable processing gain. 1. Introduction Accompanying the recent advances in digital signal processing technologies and the spread of personal comput- ers, the demand has increased for mobile multimedia com- munication capable of integrated transmission of data such as files and images in addition to conventional voice trans- mission. Since a Direct Sequence-Code Division Multiple Access (DS-CDMA) system is a wireless access technique that can flexibly handle these demands and has superior performance, such as fading performance that is robust against frequency selectivity, communication secrecy, and the ease of handoffs, developments have advanced to make this system an important access scheme for next-generation mobile communication systems [16]. In switching schemes, circuit switching is superior for media that hold a line for a relatively long time or demand immediacy such as speech or large file transmission, and packet switching is superior for relatively low capacity data transmission as in electronic mail. Consequently, a necessary condition in DS-CDMA systems is the coexistence of circuit switching and packet switching in one frequency band [79]. In packet transmission using conventional Time Di- vision Multiple Access (TDMA), when a portion or all of the packets collide, the transmissions of all of the colliding packets are lost and transmission throughput drops. Conse- ' 2001 Scripta Technica Electronics and Communications in Japan, Part 1, Vol. 84, No. 10, 2001 Translated from Denshi Joho Tsushin Gakkai Ronbunshi, Vol. J83-B, No. 1, January 2000, pp. 3948 Contract grant sponsor: Partly supported by a Grant-in-Aid for Scientific Research (B), Grant No. 11450149 from the Ministry of Education, Science, Sports and Culture. 16

Upload: takumi-ito

Post on 11-Jun-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

Wireless Packet Transmission with Variable Processing Gain

Control in Packet-Switched and Circuit-Switched Mode

Integrated DS-CDMA Systems

Takumi Ito, Seiichi Sampei, and Norihiko Morinaga

Graduate School of Engineering, Osaka University, Suita, 565-0871 Japan

SUMMARY

A DS-CDMA system with both a circuit-switching

scheme for voice transmission and a packet-switching

scheme for data transmission is effective in multimedia

communication required to transmit a variety of media.

Although the Slotted-ALOHA system is believed to be an

effective packet-switched system from the perspectives of

transmission efficiency and interference power, only open

loop transmit power control is used because of the difficulty

in using closed loop transmit power control in the ALOHA

system. In this paper, we analyze the performance of open

loop transmit power control that uses a power margin to

obtain the quality required even in packet-switched termi -

nals, and propose wireless packet transmission with vari -

able processing gain control as the method to achieve

high-speed transmission by more effectively using the free

channels. The results of computer simulations showed that

the system capacity of circuit-switched terminals is not

damaged when an appropriate power margin is set, and the

additional effect of the packet-switched terminals is in -

creased. In addition, the proposed system can greatly short -

en the average delay time of packet-switched terminals. ©

2001 Scripta Technica, Electron Comm Jpn Pt 1, 84(10):

16�26, 2001

Key words: DS-CDMA; packet switching; multi-

media communication; open loop transmit power control;

power margin; variable processing gain.

1. Introduction

Accompanying the recent advances in digital signal

processing technologies and the spread of personal comput-

ers, the demand has increased for mobile multimedia com-

munication capable of integrated transmission of data such

as files and images in addition to conventional voice trans-

mission. Since a Direct Sequence-Code Division Multiple

Access (DS-CDMA) system is a wireless access technique

that can flexibly handle these demands and has superior

performance, such as fading performance that is robust

against frequency selectivity, communication secrecy, and

the ease of handoffs, developments have advanced to make

this system an important access scheme for next-generation

mobile communication systems [1�6]. In switching

schemes, circuit switching is superior for media that hold a

line for a relatively long time or demand immediacy such

as speech or large file transmission, and packet switching

is superior for relatively low capacity data transmission as

in electronic mail. Consequently, a necessary condition in

DS-CDMA systems is the coexistence of circuit switching

and packet switching in one frequency band [7�9].

In packet transmission using conventional Time Di-

vision Multiple Access (TDMA), when a portion or all of

the packets collide, the transmissions of all of the colliding

packets are lost and transmission throughput drops. Conse-

© 2001 Scripta Technica

Electronics and Communications in Japan, Part 1, Vol. 84, No. 10, 2001Translated from Denshi Joho Tsushin Gakkai Ronbunshi, Vol. J83-B, No. 1, January 2000, pp. 39�48

Contract grant sponsor: Partly supported by a Grant-in-Aid for Scientific

Research (B), Grant No. 11450149 from the Ministry of Education,

Science, Sports and Culture.

16

quently, studies to prevent packet collisions have been

conducted on Packet Reservation Multiple Access (PRMA)

which reduces the probability of packet collisions by con-

trolling the transmission permission or Carrier Sense Mul-

tiple Access (CSMA) even under random control [10].

On the other hand, in a DS-CDMA system, all of the

channels operate in an environment where they interfere

with each other. Therefore, under the condition that an

appropriate number of terminals operate based on suitable

transmit power control, except when all of the packet-

switched terminals do not receive with the exact timing

when using the same spreading code, packets can be cap-

tured, and transmission permission control becomes useless

[11, 12]. Thus, if both circuit switching and packet switch-

ing are supported in a DS-CDMA scheme, making this

feature as practical as possible is desired.

The most significant issue when supporting packet

switching in a DS-CDMA scheme is transmit power con-

trol. Transmit power control can be broadly divided into

closed loop transmit power control that requires feedback

control at the terminals and the base station and open loop

transmit power control that estimates and compensates for

attenuation based on a pilot signal from the base station

[13�15]. A circuit-switched system transmits control infor-

mation before transmitting data and uses this control infor-

mation to enable easy implementation of closed loop

transmit power control. In contrast, a packet-switched sys-

tem transmits control information when data transmission

begins and a time equal to or longer than the time needed

to actually transmit the data is used to transmit the control

information, and the transmission efficiency drops substan-

tially. Thus, the preferred transmit power control in packet

switching only handles open loop control.

Therefore, we analyze the characteristics when the

power margin is designed to achieve transmission using

only open loop control and transmission occurs with a large

transmit power. Although the drops in the received power

caused by instantaneous fluctuations can be lessened when

a larger power margin is set, the interference powers to the

coexisting circuit-switched terminals are increased, and the

system capacity of the circuit-switched terminals is ex-

pected to fall. In contrast, the drop in capacity of the

circuit-switched terminals is avoided by designing a

smaller power margin; unfortunately, the transmission per-

formance of the packet-switched terminals degrades. Thus,

optimization of the power margin becomes an issue.

Furthermore, when there are free channels in a DS-

CDMA system with coexisting circuit-switched terminals

and packet-switched terminals, the transmission speed of

the packet-switched terminals increases and the transmis-

sion should end sooner and release the channel to other

users. This should improve the data throughput perform-

ance of the system. If the system is based on TDMA,

complex control is required such as searching for free

channels and allocating slots. However, DS-CDMA as-

sumes that the transmit power control is based on the

received power standard. The free channel situation can be

determined by the interference power at the base station and

can control the transmission speed by only controlling the

transmit power and processing gain. Transmit speed control

of a packet-switched terminal that responds to free channels

in the system can be very easily implemented.

In this paper, we propose wireless packet transmis-

sion with variable processing gain control that has the

objective of speeding up the packet-switched terminal

transmission speed by applying open loop transmit power

control for packet-switched terminals and controlling the

processing gain of the packet-switched terminals in re-

sponse to the interference power when circuit-switched

terminals and packet-switched terminals coexist. The re-

sults of analyzing the transmission characteristics of the

proposed scheme by computer simulation showed that the

proposed system could achieve high-speed packet trans-

mission without lowering the circuit capacity of the circuit-

switched terminals.

2. Packet Transmission in DS-CDMA

Systems

In a DS-CDMA system, the arrival times of the paths

must be separated by at least one chip time to separate and

combine the delayed waves, and each path must be inde-

pendent. In a mobile communication environment, when

each path is nearly independent and the target system has a

chip rate around 16 Mcps, one chip time can be regarded as

a sufficiently small time interval assuming that one chip

time is around several dozen meters when converted to a

distance difference and about 1 km is assumed as the zone

radius. Consequently, even for packets transmitted nearly

simultaneously, separation in the base station may be pos-

sible. In other words, the additional effects obtained only

when a large difference exists in the received power in a

conventional TDMA system are also obtained when no

large difference in the received power exists in a CDMA

system. This demonstrates the possibility of obtaining the

preferred result of expecting successful packet transmission

by the additional effects of actively transmitting packets

without control than by performing controls such as traffic

control to avoid simultaneous transmissions of packets in

the entire system.

Thus, in packet transmission by a DS-CDMA

scheme, random access control is applicable as an access

scheme. Because the reception characteristics of a DS-

CDMA scheme are determined by the interference power,

from the perspective of interference power, a random access

scheme is applicable only when transmitting data. The

17

random access schemes are divided into ALOHA schemes

and CSMA schemes, but the hidden terminal problem

develops in a CSMA scheme and the channel sensing is

incomplete. Considering the excellent additional effects of

a DS-CDMA system, the ALOHA system is thought to be

suitable. Furthermore, the ALOHA scheme is divided into

Pure-ALOHA and Slotted-ALOHA. Synchronized detec-

tion during reception becomes complex in Pure-ALOHA,

but Slotted-ALOHA may be applicable. Since a DS-

CDMA system operates with a margin for interference

power, a guard time does not have to be provided to prevent

packet collisions due to the transmission path charac-

teristics as in a TDMA system. Therefore, no particular

guard time is set in the analysis by computer simulation to

be described later.

3. Transmission Power Control

This section describes the transmit power control

used in DS-CDMA systems. The equal-level double-spike

Rayleigh model is assumed for the instantaneous fluctua-

tions.

Since the transmission signal fluctuates due to path

loss, short-term shadowing, and instantaneous fluctuations,

the received power Rx is given by

where A is a constant, Tx is the transmit power, r�D is the

path loss, 10[ / 10 is a stochastic variable following a loga-

rithmic normalized distribution, and J is a stochastic vari-

able representing instantaneous fluctuations. The

DS-CDMA scheme requires solving the near/far problem

produced by the distance difference between each terminal

and the base station. Thus, the most widely used control is

transmit power control. Transmit power control can be

broadly classified into closed loop transmit power control

capable of compensation that includes the instantaneous

fluctuations, and open loop transmit power control capable

of compensating only path loss and short-term shadowing.

In closed loop transmit power control, the received power

Rxc and the transmit power T x

c are given by the following

equations with R as a constant:

In the computer simulation to be described later, closed loop

transmit power control and maximum combining diversity

are used. The probability density function J Jm of the

instantaneous received power after this combining is given

by

As shown in Eq. (2), in contrast to the highly accurate power

control available when using closed loop transmit power

control, the control information needed in transmit power

control must be transmitted regardless of the presence or

absence of information that should be transmitted. Thus,

closed loop transmit power control is suited to continuous

transmission having a relatively long hold time to lessen the

transmission overhead. In the circuit-switched scheme for

voice transmission, since the transmission is considered to

be continuous, closed loop transmit power control may be

easily applied.

On the other hand, because the call hold time is

shortened and transmission is bursty in packet transmission,

closed loop transmit power control is extremely difficult to

apply and only open loop control is used. The transmit

power T xo and the received power Rx

o when using open loop

transmit power control are given by

In CDMA, all of the arrived waves having different

delay times can be separated. Since a RAKE receiver is used

for maximum ratio combining, the combining must be

performed after identifying which terminals transmitted the

waves. However, this kind of operation is believed to exces-

sively increase the control load in packet switching that

shares the spreading code, particularly during high traffic.

Thus, in the computer simulation to be described later,

selective combining diversity selects and demodulates only

one path having the maximum received power. In this case,

a significant performance improvement compared to packet

transmission is expected in TDMA due to the additional

effects. The probability density function of the instantane-

ous received power J Js is given by

As is clear from Eq. (6), when only open loop transmit

power control is used, the probability of the required re-

ceived power decreasing because the received power has

larger fluctuations than the instantaneous fluctuations is

increased, and the reception performance is expected to

degrade. Consequently, we study a technique that sets the

required received power of the packet switching terminal

beforehand to a level M times the required received power

of the circuit-switched terminal and reduces the drop

caused by instantaneous fluctuations. Thus, the increase in

the received power is defined as the power margin. The

(1)

(2)

(3)

(4)

(5)

(6)

(7)

18

transmit power T xm and the received power Rx

m are given by

the following equations with M as the power margin:

From the above, when closed loop transmit power control

and open loop transmit power control with an added power

margin are used, the received CIRs at base station 1 are

given by

where Ilc and Il

m are given by

where J and K are the number of terminals used in closed

loop control and open loop control with an added power

margin, respectively. As shown by Eqs. (2) and (9) to (13),

in contrast to the expected improvement in the performance

of terminals using open loop control with the added power

margin as the power margin becomes larger, performance

degradation is expected in the terminals using closed loop

control. Conversely, although the performance degradation

in the terminals using closed loop control is suppressed

when the power margin is decreased, the performance of

the terminals using open loop control with an added power

margin is expected to be degraded by instantaneous fluc-

tuations. Consequently, the optimum value is believed to

fall within the power margin.

4. Packet-Switched Transmission Scheme

with Variable Gain

The reception performance in DS-CDMA depends

greatly on the average interference power [3]. To satisfy the

required quality, the sum of Ilc and Il

m in Eqs. (10) and (11)

must be less than some constant value. Even if the required

received power of the packet switching terminal increases

for a small Ilc and Il

m increases, the received signal CIRs

given by Eqs. (10) and (11) satisfy the criterion. At the same

time the required received power increases for a constant

chip rate, high-speed transmission can be achieved by re-

ducing the processing gain. This algorithm is proposed

next. The sum of the interference powers at base station 1

is given by

where the subscript �own� means the terminal is in cell 1

and �other� means it is in a cell other than 1. Since the

interference power IlS when the required received power is

multiplied by 2S must be less than or equal to the constant

Ith, the largest positive integer S satisfying the following

equation is found:

In the computer simulation to be described later, the proc-

essing gain of the circuit-switched terminals is set to 256,

and the processing gain of the packet-switched terminals is

set to 128, 64, 32, and 16. As a result, S = 0 becomes

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

19

equivalent to the number of channels used by a circuit-

switched terminal during transmission. The base station

broadcasts this in the service area. The terminals transmit

with the power margin multiplied by 2S�1 and the gain

multiplied by 1 /2S�1. The transmission speed Rb at a termi-

nal is given by the following equation and the transmission

speed becomes 2S�1 faster:

where W is the chip rate, Gp is the standard processing gain,

Cr is the coding rate, and mf is the modulation level of the

primary modulation method. The transmission power T xS

is given by

Since the error rate performance is maintained when the

transmission power T xS exceeds the maximum transmission

power, S is decreased so that the transmission power falls

in the range of the maximum transmission power.

5. Computer Simulation

Table 1 lists the computer simulation parameters.

This computer simulation used 19 omnicells to perform cell

wrapping and calculated the interference power in all of the

cells from the surrounding 19 cells, including the cell itself.

The carrier frequency is set to 900 MHz, and the maximum

Doppler frequency is set to 10 Hz. This assumes that the

low-speed mobile terminal has a moving speed equivalent

to 12 km/h. Figure 1 shows the system model. In Fig. 1,

Codes 1 to n indicate the spreading codes assigned to the

circuit-switched terminals, and Code A indicates the

spreading code assigned to the packet-switched terminals.

A circuit-switched terminal for voice transmission has its

own spreading code because the call hold time becomes

longer and continuous transmission is required. In this

computer simulation, voice activity is not considered and

the hold time is set to infinity. In contrast, the packet-

switched terminals are believed to have short transmission

times, implement high-speed spreading code synchroniza-

tion, and efficiently use the spreading code. Thus, all of the

terminals share a spreading code. The traffic of packet-

switched terminals has arrival times with an exponential

distribution and message lengths with a geometric distribu-

tion [16].

Circuit-switched schemes use closed loop transmit

power control because the transmission is continuous.

Packet-switched schemes use open loop transmit power

control with an added power margin because the transmis-

sion is bursty. As shown in Fig. 1, the received power at a

(16)

Table 1. Computer simulation parameters

Cell radius 1 km

Path loss Hata formula

(large city model)

Carrier frequency 900 MHz

Antenna gain Base station 8 dB

Mobile

station

2 dB

Antenna height Base station 30 m

Mobile

station

1.5 m

Noise figure (NF) 3 dB

Standard deviation of short-term

shadowing

6.0 dB

Instantaneous fluctuations Equal-level

double-spike

Rayleigh model

Maximum Doppler frequency 10 Hz

Circuit-switched terminal Closed loop

transmit power

control

Transmit power control

Packet-switched terminal Open loop

transmit power

control

Transmit power control

Maximum transmit power 25 dBm

Minimum transmit power �65 dBm

Transmit power control step 0.5 dB

Target received power �107.7 dBm

Chip rate 16.384 Mcps

Primary modulation scheme QPSK

Processing gain

(Truth value)

Circuit 256

Packet 16, 32, 64, 128

Forward error correction Convolutional

code with 1/2

coding rate

Access protocol Slotted ALOHA

1 slot time 2 ms

Message length Geometric

distribution with a

mean of 100 bytes

Arrival time Exponential

distribution with a

mean of 80 slots

(17)

20

circuit-switched terminal is nearly constant. As explained

in Section 3, however, the received power fluctuates greatly

in a packet-switched terminal because the received power

is varied by fading fluctuations.

In this paper, one chip time is fixed regardless of the

processing gain used in the transmission because the simu-

lation is performed with a fixed chip rate. Thus, for simplic-

ity, all of the waves arriving at the base station are separated

by at least one chip time on arrival.

The required quality of the circuit-switched terminal

in this computer simulation is set to a 1% or lower outage

probability. The outage probability is defined as the prob-

ability of the bit error rate exceeding 10�3. A system capac-

ity of 35 circuit-switched terminals per cell for this system

is verified by computer simulation.

5.1. Power margin optimization

Since one spreading code is shared by all terminals

in a packet switching scheme, if the input traffic illustrating

the level of transmission requests exceeds 1 erl/cell, the

delay performance degrades substantially. Therefore, we

consider 0.08 erl/cell input traffic as low traffic and 0.8

erl/cell input traffic as high traffic. Figure 2 shows the

normalized delay performance of the packet-switched ter-

minals. Figure 3 shows the outage probability of circuit-

switched terminals. Here, the delay time is defined as the

time needed until each packet is correctly received from call

origination by means of ARQ. The normalized delay is

defined as the delay time normalized by the delay time for

transmission using an error-free channel at a processing

gain of 128.

Figure 2 shows the substantial improvement in the

delay performance of the packet-switched terminals as the

power margin increases. The reason may be the higher

probability of being able to compensate for the degradation

caused by instantaneous fluctuations as the power margin

is set to a larger value. Favorable performance is obtained

regardless of the number of circuit-switched terminals in a

cell when the power margin is set to at least 5 dB, that is,

the interference power. When open loop transmit power

control and two-path selective combining diversity are

used, the probability of the received power dropping by

more than 5 dB is 10% or less [17]. Thus, by setting the

power margin to 5 dB, the drop caused by instantaneous

fluctuations can be decreased at a probability of 90% or

higher.

Figure 3 clearly shows degradation in the outage

probability of circuit-switched terminals as the power mar-

gin is set to a larger value. This may be due to the increase

in the interference power on the circuit-switched terminals

because the packet-switched terminals are set with a larger

power margin. From Fig. 3, the performance of the circuit-

switched terminals degrades particularly for 30 terminals

and 0.8 erl/cell when the power margin is set to a value

Fig. 1. System model.

Fig. 2. Power margin versus normalized delay

performance.

Fig. 3. Power margin versus outage probability.

21

larger than 7 dB. The outage probability of 1%, which is

the required quality, is not satisfied. Setting the power

margin to a value larger than 7 dB by the packet-switched

terminals is believed to be equivalent to a circuit using at

least five lines for transmission. Consequently, the perform-

ance of circuit-switched terminals degrades because of the

higher probability that a terminal instantaneously exceed-

ing the system capacity is in the cell.

When the power margin is set from 5 to 7 dB, ade-

quate improvement in the delay performance is obtained for

packet-switched terminals without introducing major per-

formance degradation in the circuit-switched terminals.

Thus, the optimum power margin is 5 dB in the following

analysis.

5.2. Comparison with open loop control

Figure 4 shows the normalized delay performance of

packet-switched terminals when there are 10 circuit-

switched terminals. Figure 5 shows the normalized delay

performance of packet-switched terminals when there are

30 circuit-switched terminals. Figure 6 shows the outage

probability performance of circuit-switched terminals for

10 and 30 circuit-switched terminals. The power margin of

5 dB obtained in Section 5.1 is used. For comparison, the

performance when open loop transmit power control with-

out a set power margin and the performance when complete

closed loop transmit power control can be achieved tempo-

rarily in packet transmission are both shown.

From Figs. 4 and 5, the performance when the power

margin was set to 5 dB is excellent compared to when a

power margin was not used. On the other hand, the perform-

ance when using closed loop control degrades dramatically

for 30 circuit-switched terminals in a cell in contrast to the

good performance for 10 terminals. In contrast to the 5-dB

improvement in the average received carrier-to-interference

ratio (CIR) when the power margin was set to 5 dB, the

additional effect of the packets diminishes due to the

smaller variance of the received CIR when using closed

loop control. In particular, when the number of circuit-

switched terminals becomes large and the interference

power is large, the number of retransmissions increases.

Transmission with the added 5-dB power margin is

expected to increase the interference power on the circuit-

switched terminals because this transmission is equivalent

to using the spectral resource of about three channels at

most to transmit one channel. As is clear from Fig. 6, when

the packet-switched terminals transmit with an added 5-dB

power margin, a 1% outage probability, which is the re-

quired quality, is satisfied although some degradation is

evident in the outage probability performance of the circuit-

switched terminals.

Fig. 4. Normalized delay performance for 10 or fewer

circuit-switched terminals.

Fig. 5. Normalized delay performance for 30 or fewer

circuit-switched terminals.

Fig. 6. Outage probability performance.

22

From the above, the application of open loop transmit

power control provided with a power margin can reduce the

drop in the received power and obtains a substantial im-

provement in performance compared to using open loop

transmit power control while inviting only slight degrada-

tion in the performance of the coexisting circuit-switched

terminals. In particular, a major improvement is obtained

with the system in a state near saturation.

5.3. Performance when using wireless packet

transmission with variable processing

gain control

The delay performance in Fig. 2 shows that when the

power margin was set to 5 dB, the delay time became nearly

constant regardless of the number of circuit-switched ter-

minals. A free channel is available when the system has a

small number, such as 10 or 20, circuit-switched terminals

in a cell. Thus, when this channel is used in transmission

by packet-switched terminals, the delay performance of the

packet-switched terminals can be expected to improve

without degrading the transmission quality of the circuit-

switched terminals.

The performance of the proposed scheme depends

greatly on Ith in Eq. (15). Setting a large Ith improves the

transmission performance of packet-switched terminals be-

cause the larger setting is equivalent to increasing the

number of channels assigned to the packet-switched termi-

nals. Simultaneously, however, the interference power on

the circuit-switched terminals increases. In contrast, setting

a small Ith does not increase the interference power, but the

spectral resource cannot be effectively employed in the

system because the number of channels allocated to the

packet-switched terminals is small. Consequently, Ith must

be optimized. In a CDMA system, operation can be close

to the interference power limit, and the wireless resources

can be effectively used.

When closed loop transmit power control was used,

the received Eb / �N0 � I0� caused by control errors follows

a normalized logarithmic distribution having a variance

around 1 to 2 dB [14], and the Eb / �N0 � I0� required to reach

a bit error rate of 10�3 is about 3 dB. To keep the outage

probability of circuit-switched terminals to 1% or less, the

average received Eb / �N0 � I0� must be about 5 to 10 dB. In

this paper, as shown in Table 1, since the target received

power is set to �107.7 dBm and the processing gain to 256,

the optimum Ith is believed to be close to �94 to �89 dBm.

Figure 7 shows the outage probability of the circuit-

switched terminals versus Ith. Figure 8 shows the normal-

ized delay performance of the packet-switched terminals.

Figure 7 reveals that when Ith exceeds �93 dBm, the per-

formance of the circuit-switched terminals degrades par-

ticularly when there are 30 circuit-switched terminals per

cell. This occurs because the interference power on circuit-

switched terminals increases due to excess use of the chan-

nels by the packet-switched terminals. As a result, the

required quality is not achieved.

On the other hand, Fig. 8 shows that the performance

of the packet-switched terminals becomes excellent as Ith is

increased. From Fig. 8, when Ith is set to �93 dBm, the

performance nearly coincides except for 30 circuit-

switched terminals per cell. When the traffic of the packet-

switched terminals is high, the unused resource in the

system is efficiently allocated to the packet-switched termi-

nals, and improvements in the delay performance are de-

signed. The delay performance is improved even when

there are 30 circuit-switched terminals and the system is in

a state close to saturation. Therefore, Ith = �93 dBm is used

in the following analysis.

Figures 9 and 10 show the delay performance and

outage probability when using the proposed scheme. This

is shown with the power margin fixed at the 5 dB obtained

Fig. 7. Ith versus outage probability performance.

Fig. 8. Ith versus normalized delay performance.

23

in Section 5.1 as the conventional method. From Fig. 9, the

delay performance of the packet-switched terminals is im-

proved by using the proposed scheme. First, when the

number of circuit-switched terminals is small (10) and there

is a large number of free channels in the system, an ex-

tremely large improvement is obtained by applying a small

processing gain such as 16 and 32. In particular, the nor-

malized delay time of 12.6 for input traffic of 0.8 erl/cell is

reduced to about 20% at 2.2. Even if the number of circuit-

switched terminals is 30, the delay time is reduced by using

a processing gain of 64 for a few free channels. For exam-

ple, for 0.8 erl/cell input traffic, the normalized delay time

of 16.2 is reduced to about 80% at 12.6.

Figure 10 shows that a particularly large degradation

is not seen in the outage probability of the circuit-switched

terminals. This has very little effect on circuit-switched

terminals when using the proposed scheme because packet-

switched terminals effectively use the free channels avail-

able in the system and not the channels allocated to

circuit-switched terminals.

Thus, the proposed scheme more effectively uses the

spectral resource and does not have a major effect on the

coexisting circuit-switched terminals. In particular, if the

number of circuit-switched terminals is 10, the average

delay time can be reduced to about 20%.

6. Conclusion

We have constructed a system with a circuit switching

scheme coexisting with a packet switching scheme in DS-

CDMA, analyzed the performance of open loop transmit

power control using the power margin, and proposed wire-

less packet transmission with variable processing gain con-

trol based on the interference power. The performance of

the proposed scheme was analyzed by computer simula-

tions.

The result was the ability to substantially reduce the

drop in the received level by increasing the power by 5 dB

and transmitting when open loop transmit power control is

used. Clearly, using the proposed scheme does not have a

significant effect on the outage probability performance of

the circuit-switched terminals and can greatly reduce the

average delay time of packet-switched terminals. In par-

ticular, the delay time can be reduced to about 1/5 for 10

circuit-switched terminals and a 0.8 erl/cell input traffic of

the packet-switched terminals.

Acknowledgments. We are very grateful to Mr.

Masato Tanaka (Uniden) for his cooperation in writing this

paper. A portion of this research is supported by a Grant-in-

Aid for Scientific Research (B), Grant No. 11450149 from

the Ministry of Education, Science, Sports and Culture.

REFERENCES

1. Gilhousen KS, Jacobs IM, Padovani R, Viterbi AJ,

Weaver LA, C.E.W. III. On the capacity of a cellular

CDMA system. IEEE Trans Veh Technol

1991;40:303�311.

2. Jung P, Baier PW, Steil A. Advantages of CDMA and

spread spectrum techniques over FDMA and TDMA

in cellular mobile radio applications. IEEE Trans Veh

Technol 1993;42:357�364.

3. Kohno R, Meidan R, Milstein LB. Spread spectrum

access methods for wireless communications. IEEE

Commun Mag, p 58�67, January 1995.

Fig. 9. Normalized delay performance with the

proposed scheme.

Fig. 10. Outage probability performance with the

proposed scheme.

24

4. Newson P, Heath MR. The capacity of a spread

spectrum CDMA system for cellular mobile radio

with consideration of system imperfections. IEEE J

Sel Areas Commun 1994;12(4).

5. Adachi F, Sawahashi M. Wideband wireless access

based on DS-CDMA. IEICE Trans Commun

1998;E81-B:1305�1316.

6. Tanaka S, Sawahashi M, Adachi F. Pilot symbol-as-

sisted decision-directed coherent adaptive array di-

versity for DS-CDMA mobile radio reverse link.

IEICE Trans Commun 1998;E80-A:2445�2454.

7. Pursley MB. The role of spread spectrum in packet

radio networks. Proc IEEE 1987;75(1).

8. Manji S, Zhuang W. Capacity analysis of an inte-

grated voice/data DS-CDMA network. ICC�97,

Montreal, Vol. 2, p 979�983.

9. Nagatsuka M, Ishikawa Y, Uebayashi S. Data traffic

control and capacity evaluations for voice/data inte-

grated transmission in DS-CDMA. IEICE Trans

Commun 1998;E81-B:1355�1364.

10. Nanda S, Goodman D, Timor U. Performance of

PRMA: A packet voice protocol for cellular systems.

IEEE Trans Veh Technol 1993;42:78�86.

11. Hata M, Yamauchi Y. Packet radio networks. CQ

Publishing; 1990. (in Japanese)

12. Ogawa A. CDMA and next generation mobile com-

munication systems. Triceps Publishing; 1996. (in

Japanese)

13. Zander J. Performance of optimum transmitter power

control in cellular radio systems. IEEE Trans Veh

Technol 1992;41:57�62.

14. Viterbi AM, Viterbi AJ. Erlang capacity of a power

controlled CDMA system. IEEE J Sel Areas Com-

mun 1993;11(6).

15. Kudoh E, Okazaki I, Ogose S. The lower bound of

open loop transmission power control error in DS-

CDMA systems. IEICE Trans Commun 1997;E80-

B:1805�1809.

16. Karlsson M, Guerin N, Laaksonen N, Ostermayer G.

Evaluation of handover algorithms for packet trans-

mission in WCDMA. VTC�99, Houston, Vol. 2, p

1402�1406.

17. Okumura Y, Shinji M. Foundations of mobile com-

munication. IEICE; 1986. (in Japanese)

AUTHORS (from left to right)

Takumi Ito (student member) received his bachelor�s and master�s degrees in engineering from Osaka University in 1997

and 1999. He is now in the doctoral program and conducts research on digital mobile communication systems. He is a student

member of IEEE.

Seiichi Sampei (member) received his bachelor�s and master�s degrees in engineering from Tokyo Institute of Technology

in 1980 and 1982. He then joined the Radio Research Laboratory, now the Communications Research Laboratory, of the Ministry

of Posts and Telecommunications. He has performed research on fading compensation, interference wave compensation, and

highly efficient modulation schemes for digital land-mobile communication. From 1990 to 1991, he was a visiting researcher

at the University of California, Davis. In 1993, he became an associate professor in the Faculty of Engineering, Osaka University.

He is the recipient of the 1985 Shinohara Young Engineer Award and the 1992 Telecom System Technology Award. He is a

member of the Institute of Television Engineers of Japan and IEEE.

25

AUTHORS (continued)

Norihiko Morinaga (member) received his bachelor�s degree in electrical engineering from Shizuoka University in 1963

and his doctorate from Osaka University in 1968. He became a research assistant, lecturer, and associate professor, and is now

a professor in the Faculty of Engineering, Osaka University. His research concerns wireless, optical, satellite, and mobile

communication systems, and EMC. From 1998 to 1999, he was president of the Telecom Society of IEICE. He is now

vice-president of IEICE. He received the 1987 Telecom Natural Science Award, the 1993 Telecom System Technology Award

from the Telecommunications Advancement Foundation, and the 1995 IEICE Excellent Paper Award. He is the translator of

Optical Communication Systems and a coauthor of Optical Communication Theory and Applications. He is a member of the

Institute of Television Engineers of Japan and a senior member of IEEE.

26