SOFT HANDOFF IN CDMA SYSTEMS

Krishna B Rao, Lakshmi Narayan Mishra

E-mail: braos@utdallas.edu, mishra@utdallas.edu

Abstract

This report analyzes the performance of a soft handoff algorithm of

the type that has been proposed in the IS-95 CDMA standard. In the soft

handoff scheme, multiple base stations are involved in the

communication to and from one mobile station to improve the handoff

performance on the boundaries between two base stations by providing

channel diversity. On the other hand, there are additional resources used

during soft handoff. There is, thus, a tradeoff between diversity

advantage and resource utilization. We present a simulation that

provides quantification for this tradeoff. The results can be used to gain

insight and help select the appropriate handoff thresholds.

The act of transferring support of a mobile from one base station to

another is termed handoff. Handoff occurs when a call has to be handed

off from one cell to another as the user moves between cells. In a

traditional "hard" handoff, the connection to the current cell is broken,

and then the connection to the new cell is made. This is known as a "

break-before-make" handoff. Since all cells in CDMA use the same

frequency, it is possible to make the connection to the new cell before

leaving the current cell. This is known as a "make-before-break" or "soft"

handoff. Soft handoff requires less power, which reduces interference

and increases capacity. The implementation of handoff is different

between the narrow band and the CDMA standards.

This is an effort to analyze a soft handoff strategy, taking into

consideration the trade-offs between selecting system parameters and

the performance benefits. We consider a scenario involving two base

stations and a single mobile, where the changes in the pilot signal

strength are tracked and a conditional decision is made allowing

communication with only one base station. This results as an outcome of

simple measurements, which determine the strongest signal and results

in handover to the corresponding base station.

In a CDMA system the same frequency band is shared between all

the cells. Thus there is well-defined efficient bandwidth utilization.

Though there is frequency reuse the orthogonal nature of the waveforms

band. A typical handoff scenario is taken into account wherein the need

for a mobile to alter its frequency or rather to switch its carrier frequency

is negated. The main idea of a soft handoff scheme is to ensure that there

is connectivity with the old base station while the new base station has

been assigned to take control over the communication link. This way at a

given instant of time we have the mobile maintaining a constant

communication link with at least one base station simultaneously

ensuring a non disrupted call activity. The algorithm may be designed in

such a way as to ensure that as soon as the mobile is within the range of

the new cell, the old base station releases the connection of the call.

We have analyzed the performance of an existing soft handoff

algorithm with the help of an analytical model. The outcome of such an

experiment utilized certain parameters such as the number of Active Set

updates, the assignment probabilities and the outage probabilities and

the number of base stations with respect to a single mobile. The

algorithm and the results obtained from the simulation are discussed in

the next few sections.

The Importance Of Soft Handoff

In power controlled CDMA systems soft handoff is preferred over

hard handoff strategies. This is more pronounced when the IS-95

standard is considered wherein the transmitter [the base station] power

is adjusted dynamically during the operation. Here the power control and
soft handoff are used as means of interference-reduction, which is the

primary concern of such an advanced communication system. The

previous and the new wideband channels occupy the same frequency

band in order to make an efficient use of bandwidth, which makes the

use of soft handoff very important. The primary aim is to maintain a

continuous link with the strongest signal base station otherwise a

positive power control feedback would result in system problems. Soft

handoff ensures a continuous link to the base station from which the

strongest signal is issued.

Problem Definition

The main function of the soft handoff algorithm is to maintain an

Active Set based on the measured pilot strength. We have the forward

traffic channel and the reverse traffic channel. A reference signal is

considered as the pilot signal referring to the forward CDMA traffic

channel. There are several such pilot signals in the channel. The main

function of the mobile station is to measure and report the pilot signal

strengths to the base station. This is a repetitive process. As soon as the

measured signal crosses a certain predefined threshold value it becomes

a member of the Active Set. The main problem lies in identifying this

threshold level. A reference pilot is picked from the Active Set and then a

handoff decision is made after comparing the received pilot strength with

that of the reference. Originally the Active Set consists of a single base
station but after repeated experiments and measurements other base

stations are added when the signal strength between them and the user

exceeds the predefined add threshold. There are three parameters that

are considered in the analysis - Tadd or the Add Threshold, Tdrop or the

Drop Threshold and the drop timer. Both Tadd and Tdrop are negative

values. If any non-member pilot signal strength exceeds the Tadd then the

pilot is added to the set or is made a member of the set. The pilot is still

continuously measured and at any instant if it drops below the drop

threshold, Tdrop, the mobile starts a drop timer. This timer is designed

such that it is reset and disabled as soon as the pilot strength goes above

Tdrop before the drop timer expires. Upon expiry of the timer the pilot also

is removed from the Active Set. The experiment was conducted for

several values and it was found that high thresholds and long drop timer

settings tend to maintain more stations in the Active Set. Similarly low

thresholds and short drop timer settings increased the rate of Active Set

updates. Here the base station utilization factor and the number of

Active Set updates serve as a good yardstick to measure the system

performance as is evident from the results obtained.

Software Implementation

The simulation of the algorithm described above in MATLAB using

a single program. We considered an environment in which we had two

base stations A & B and single mobile user moving from base station A in
the direction of base station B with no interference. A pilot signal is

assumed to be transmitted by each base station that is measured at the

mobile and used for all decisions. The pilot signal from each base station

S a (d ) = K1 − K 2 x log(d ) + u (d )
is modeled by the formula: where d is the
Sb (d ) = K 1 − K 2 x log(d ) + v(d )

distance, K1 and K2 are constants that can be used to model any given

environment and u and v are gaussian random noise sequences.

The main parameter of interest, namely, the Active Set is basically

a registry that lists base stations that are capable of providing reliable

communication with the mobile. Reliable communication refers to the

threshold levels that the received pilot signal must satisfy and these

thresholds are programmable by the user. The Active Set also holds the

base station that the mobile is currently communicating with and in case

a handoff is required, a base station from the Active Set is chosen as the

next base station. The user can also choose drop timer values and the

number of sampling points as well as the number of simulation cycles.

The program provides the user the option to choose from any of

two simulations. The first simulation plots the assignment probabilities

to base stations A & B and the probability that both base stations A & B

are in the Active Set as well as the Outage Probability (no base station in

the Active Set and hence call dropping) with respect to the distance. The

second simulation plots the number of Active Set updates against the

size of the Active Set for varying values of the drop timer.
 Results

We provide results of the above-mentioned simulations with the

optimum values of all parameters obtained by running this simulation

over many sets and combinations of values.

As expected the assignment probability of base station A decreases

with increasing distance. This is because we have assumed the user to

be moving from base station A in the direction of base station B.

Similarly the assignment probability of base station B increases with

distance. The probability of both the base stations being in the Active Set

is maximum at the boundary region of the two cells. This is expected

because the pilot signals have almost the same strength in this region. It

is interesting to note that the outage probability is also maximum in the

boundary region. The inference we can draw from this scenario is that an

effort to reduce the outage probability by manipulating the threshold

values (mostly decreasing the thresholds) will result in more base

stations existing in the Active Set. Also the width of the graph in figure 2

is increased (along with the height of the peak). This implies that overall

there will be more base stations for a longer period of time in the Active

Set thus requiring more system resources to implement the algorithm.

Therefore a tradeoff has to be struck between the outage probability and

the system resources required.

the number of base stations in the Active Set with increasing drop timer

values. It can be seen that there is a sharp decrease in number of

updates from a timer value of 1 to 5. Also increasing the timer value any

further does not produce any significant improvement in the

performance. The number of updates represents the system overhead

incurred in this algorithm. Hence it can be inferred that a drop timer

value of 5 is ideal for this algorithm.

Conclusions

We have presented a very simple and efficient Soft Handoff

algorithm. The performance of the soft handoff algorithm can be analyzed

using the simulation we have presented. Our conclusion is that the

introduction of the drop timer greatly reduces the system overhead with

only a slight increase in the resource usage. As a next step the same

algorithm could be extended to accommodate more base stations and

interference factor could also be considered. It could be an interesting

feature to observe the behavior of this model in the event of non-

availability of the channel.

[1] Ning Zhang and Jack M. Holtzman, “Analysis of a CDMA Soft Handoff
Algorithm”, IEEE Trans. on Vehicular Technology, vol.47, no.2, May 1998,
pp.710 – 714.
[2] Daniel Wong and Teng Joon Lim, “Soft Handoffs in CDMA Mobile Systems”,
IEEE Trans. on Personal Communications, vol. 4, no. 6, Dec 1997, pp. 6 – 17.
[3] Theodore S. Rappaport, “Wireless Communications – Principles & Practices”,
Prentice Hall, NJ, 1996.
[4] Suwon Park, Ho Shin Cho, Dan Keun Sung, “Modelling and Analysis of
CDMA Soft Handoff”, Proc. IEEE VTC, vol. 3, 28 April – 1 May 1996, pp. 1525
– 29.
[5] http://www.cdg.org
[6] http://www.mathworks.com