A Traffic Shedding Algorithm for Soft-Handoff in MC-CDMA Systems
Bih-Hwang Lee Hsin-Pei Chen Su-Shun Huang
Department of Electrical Engineering
National Taiwan University of Science and Technology
lee@ccg.ee.ntust.edu.tw
Abstract
In mobile communication environment, an abrupt increase of mobile users in a specific cell will make traffic overloaded and become a so-called hot-spot cell, which causes an unbalanced traffic load condition because the traffic load in the neighboring cells may be light. In order to utilize a limited bandwidth efficiently, it is necessary to use a good mobile handoff strategy to alleviate the unbalanced traffic condition. This paper proposes a complete traffic shedding algorithm (CTSA) for soft handoff efficiently to alleviate the overloaded traffic in the hot-spot cell by adjusting handoff parameters, which virtually reduces the service coverage of the heavy loaded cell without adjusting the transmission power. The mobile users located in the outer area of the hot-spot cell will early handoff to the neighboring cells with light traffic load; and the mobile users in the neighboring cells postpone to handoff to the hot-spot cell, so that the traffic load of the hot-spot cell can be decreased. The proposed algorithm is analyzed and evaluated by simulation; CTSA algorithm shows better performance than the others; it obviously increases system throughput and decreases soft handoff call dropping ratio, and new call blocking ratio. Keyword: mobile communication, hot-spot cell, traffic shedding, soft handoff.
1. Introduction
Wireless communication techniques have rapidly been developing recently. In order efficiently to utilize the limited bandwidth, a variety of handoff methods are proposed to achieve steady communication during handoff procedure. In the handoff procedure of traditional CDMA, pilot signal is compared with the fixed threshold parameters, which are defined at the beginning of system startup and cannot be dynamically adjusted for the appropriate values according to the traffic load [1]. For instance, a huge traffic may be aggregated into a cell to form a hot-spot cell when
many people suddenly enter the cell at the same time for a specific activity. It may cause an unbalanced traffic compared with the other neighboring cells, which may also increase the burden of base station (BS) and cause the failures of connecting new calls or on-going calls due to the lack of bandwidth [2].
In order to resolve the above-mentioned problems, some related researches have been proposed, such as dynamical channel allocation [3], changing service region by adjusting pilot signal power [4][5], dynamical adjustment handoff threshold parameters [2], [6][8] etc.. The method of changing service region by adjusting pilot signal power is to use the characteristics of power control and the soft handoff procedure in CDMA system. It shrinks the service region of hot-spot cell by reducing the power of BS’s pilot signal, hence it is also called cell shrinking scheme. However, this method easily leads to service holes and increases the probability of handoff failure, because BS cannot independently operate with another BS and rapidly react for the change of traffic load.
For the method of dynamical adjustment handoff parameters, it is simpler and feasible for BS to perform this operation independently, which mainly includes the variable threshold soft handoff (VTSH) and enhanced soft handoff (ESH) methods [7][8]. VTSH cannot perform very well especially for the heavy traffic load because it adjusts the handoff parameters only once. ESH frequently changes handoff parameters as hot-spot phenomenon occurs, because it does not define the hysteresis parameter, which easily causes the ping-pong effect of handoff procedure and increases the system burden.
This paper proposes a complete traffic shedding algorithm (CTSA) to solve the unbalanced traffic load problem without adjusting the transmission power of BS’s pilot signal. CTSA can effectively adjust handoff parameters dynamically depending on the traffic load variation. The mobile users located in the hot-spot cell boundary will early handoff to a neighboring cell having light traffic load; conversely, the mobile users
located in the neighboring cells are postponed to handoff to the hot-spot cell. BS can independently accomplish the traffic balancing without adjusting the pilot signal power of the neighboring cells.
2. Soft handoff in MC-CDMA systems
This section briefly explains the traditional code division multiple access (CDMA) handoff mechanism. CDMA uses direct sequence spreading spectrum (DSSS) technique to transmit data. In order to satisfy the bandwidth requirements for different transmission rates, multi-carrier CDMA (MC-CDMA) can be used [9][10]. The advantages of MC-CDMA technique are to increase throughput, resist noise and interference, and decrease bit error rate, so it is suitable to be used in wireless environment with the limited bandwidth.
There are four handoff parameters related to the measurement of pilot signal for the basic handoff procedure in CDMA [1], [7], which are T_ADD, T_COMP, and T_DROP and T_TDROP. T_ADD is to indicate pilot detection threshold; a mobile terminal (MT) may send a pilot strength measurement message (PSMM) to BS if MT detects that the pilot strength is greater than T_ADD. The PSMM contains the following information for each of the pilot signals: estimated Ec/It, arrival time, and handoff drop timer.
T_COMP is to indicate comparison threshold; BS may send a handoff direction message (HDM) to MT when BS receives the pilot strength of PSMM to be greater than the sum of T_ADD and T_COMP. T_DROP and T_TDROP are to indicate pilot drop threshold and drop timer threshold, respectively; MT starts the T_TDROP timer if the pilot strength falls below the T_DROP value; and MT must leave the BS’s service region if the timer expires. Figure 1 shows the basic handoff procedure in CDMA that an MT moves from base station A (BSA) toward base station B (BSB) [1].
For traditional soft handoff (CSH) method, the handoff region is always fixed, so the handoff parameters cannot be changed to follow the traffic load variation. Mobile is subject to higher refusing ratio because of the insufficient resources, when the hot-spot problem occurs.
3. Traffic shedding procedure
In order to resolve the hot-spot problem, the concept of traffic shedding can be applied that the mobiles in hot-spot cell should early handoff to the neighboring cells having light traffic load; and the mobiles in the neighboring cells should postpone to handoff into the hot-spot cell. Figure 2 shows the concept of changing handoff parameters when traffic shedding starts. We assume that BSA is a hot-spot cell and an MT is leaving
Figure 1 The basic handoff procedure in CDMA
Figure 2 The concept of changing handoff parameters on traffic shedding
BSA for BSB. The soft-handoff region can be shifted
closer to the hard-handoff boundary of BSA, if the
T_ADD value of the entering BSB decreases and the
T_DROP value of the leaving BSA increases
[11][12]. When shedding mechanism starts to execute, BS must adjust the handoff parameters, e.g., T_ADDj and T_DROPj are the T_ADD and T_DROP
values after the jth traffic shedding, respectively. Therefore the mobile can early handoff to the neighboring cell and release the occupied resource from BSA.
It is necessary for a recovery mechanism after performing the traffic shedding if the traffic load of the hot-spot cell becomes lighter later, i.e., T_ADD and T_DROP should be resumed to the previous values. The recovery mechanism makes the mobile go back the normal handoff procedure, prevent from causing unnecessary early handoff to the neighboring hot-spot cell, and alleviate the congestion status in the neighboring hot-spot cell.
3.1. Relationship between handoff region and shedding ratio (SR)
Shedding ratio (SR) concerns the adjusted variation of the threshold values and the related service region when traffic shedding occurs. Figure 3 shows the relationship between handoff region and shedding ratio. Before explaining the relationship between handoff region and shedding ratio, we assume that mobiles are uniformly dispersed in each service region, the traffic density in hot-spot cell is greater than the other neighboring cells, and the original cell radius and hard-handoff boundaries are R0 and Rh, respectively, as
shown in Fig. 3. In order to analyze the proposed algorithm, several system parameters are defined as follows.ȡ is the bandwidth utilization of BS, which is defined as the used bandwidth divided by the total bandwidth of BS. D and Ȗ are the threshold values of traffic loads for traffic shedding and recovery mechanisms, respectively, where the shedding and recovery mechanisms start when ȡ > Į and ȡ < Ȗ, respectively. ȕ is the upper limit value of traffic load after adjusting and set between Į and Ȗ; ȡ is adjusted from Į to ȕ when shedding mechanism starts. K is a flag to indicate if the shedding or recovery mechanism starts, e.g., it is set to 1 for starting the shedding or recovery mechanism and 0 for ending the recovery mechanism. rj is the cell service radius after the jth
shedding. Tdur is the duration that BS checks the traffic
load if the recovery mechanism should start.
Figure 3 The relationship between handoff region and shedding ratio
The shedding ratio after the jth shedding, SRj, can be
derived as shown in Eqs. (1)(2):
SR1 = R0 – r1 (1)
…
SRj = rj-1 – rj (2)
Let us further denoteȡm to be the mobile density in
each service region;Di,Eiandȡmi (for i = 1, 2, …, j) are
the values ofD, E and ȡm at the ith shedding, obtained
step-by-step until the jth shedding. According to the previous mentioned assumptions, the relations among Di,Eiand ri can be shown as in Eqs. (3)(4).
2 0 1 2 0 1 2 1 1 1 1 ¸¸ ¹ · ¨¨ © § R r R r m m S U S U D E (3) …. 2 1 2 1 2 ¸ ¸ ¹ · ¨ ¨ © § j j j mj j mj j j r r r r S U S U D E (4)
Let us further assume thatDiandEi (for i = 1, 2, …, j)
have the same values and are equal to D and E, respectively, as shown in Eqs. (5)(6). Finally, rj and
SRj can be obtained by simplifying Eqs. (3)(4) as
shown in Eqs. (7)(8), respectively.
D1 = D2 = = Dj = D (5) E1 = E2 = = Ej = E (6) 0 2 j 1 j 2 1 j r R r ¸ ¹ · ¨ © § ¸ ¹ · ¨ © § D E D E (7) 0 2 1 2 1 1 R SR j j u » » » ¼ º « « « ¬ ª ¸ ¹ · ¨ © § u ¸ ¹ · ¨ © § D E D E (8) 3.2. CTSA algorithm
This subsection explains the main processes of CTSA, including traffic shedding, recovery, and postponing algorithms.
(1) Traffic shedding algorithm
Step 1: If the traffic load exceeds the threshold valueĮ, BS starts the shedding mechanism. Step 2: BS decides the shrunk service radius of
handoff region and sets flagȘ to be 1. Step 3: The service radius is set to be the hard
handoff boundary Rh if the shrunk service
radius is less than Rh.
Step 4: BS broadcasts the corresponding handoff threshold parameters to mobiles for updating.
(2) Recovery algorithm
Step 1: BS periodically checks the traffic load in the duration Tdur. Recovery mechanism
starts if flag Ș is set to 1 and the traffic load is less than the thresholdȖ.
Step 2: BS resumes the service region according to the previous shedding ratio.
Step 3: FlagȘ is set to 0 if the service radius of BS reaches the original cell boundary R0 after
adjustment.
parameters to mobiles for updating after adjustment.
(3) Postponing algorithm
A mobile sends a PSMM to a new neighboring BS if the mobile detects that the pilot strength of the new BS is greater than the threshold T_ADD+ T_COMP. The PSMM message includes the information of handoff drop timer to indicate how much time the mobile will be dropped by the original BS [1], therefore, the new BS can use this information to control when the mobile should be admitted to prevent on-going mobiles from being dropped. Figure 4 shows the timing sequence on the threshold when BSB is a hot-spot cell and a mobile
is in the service region of BSA and goes toward the
service region of BSB. A mobile normally sends a
PSMM to BSB again at timestamp (4), when the
pilot strength of BSB is greater than
T_ADD+T_COMP. BSB then admits the mobile if
BSB has enough resource. If BSB is in the hot-spot
status, we can control the timing of sending HDM to postpone the mobile entering the service region of BSB by cooperating the handoff drop timer of
PSMM and the resource of BSB, where the
postponed interval is between timestamps (5) and (7). That is, at the timestamp (5), the mobile sends a PSMM to BSB with the value of handoff drop timer
to be greater than zero. BSB admits the mobile if
BSB has enough resource for the mobile request.
Otherwise, BSB keeps postponing the mobile
entering until the handoff drop timer expires at timestamp (7). Therefore, BSB does not temporarily
admit the mobile to save resource if the pilot strength is between timestamps (4) and (5).
Figure 4 The timing sequence on the threshold
when BSB is a hot-spot cell
3.3. Relationship between service radius and handoff parameters
The process of mobile handoff is either to leave or enter the service regions of the original BS or another BS, respectively. Generally, a mobile may receive a pilot signal from BS; Li(di,] i) denote the strength of the
pilot signal sent by BSi, where di and ] iare the distance
and the shadowing fading effect between the mobile and BSi, respectively. Li(di,] i) can be obtained by
propagation model and shown as Eq. (9) [6], [13]:
,
1010 i m i i i i d pd L ] ] ʿʳ (9)
where p is the transmitter signal power of BSi which is
a constant value; m consists of path loss component and the effect of multi-path fading. Let us further assume that ] i is Gaussian distribution with zero mean and
standard deviation ı. The log-normal form of Eq. (9) can be expressed as Eq. (10):
i i
dB i i
i d p m d
L ,] 10log 10 log ] (10) We assume that BSB is a general cell but BSA is a
hot-spot cell and runs the shedding algorithm to change the handoff threshold. A mobile will be postponed to enter the service region of BSA if it is moving from BSB
to BSA. Conversely, in the opposite direction, the
mobile will handoff to BSB earlier to alleviate the
congestion situation. Let us denote that LA(rj,]A) and
LB(rj,]B) are the strengths of pilot signals sent by BSA
and BSB, respectively, which satisfy the conditions of
Eqs. (11)(12) according to the shedding algorithm if the mobile in the service region of BSA is approaching
the service region of BSB.
LB(rj,]B) >T_ADDj+T_COMP = T_ADD–
'T_ADDj + T_COMP (11)
LA(rj,]A) >T_DROPj = T_DROP+'T_DROPj (12)
'T_ADDjand'T_DROPj are the adjustment ranges
of handoff threshold after the jth shedding. From Eqs. (8) and (11),'T_DROPj can be obtained as follows.
'T_DROPj = T_DROPj T_DROP
= LA(rj,]A) LA(R0,] A) ¸ ¹ · ¨ © § ¸ ¹ · ¨ © § ¸ ¸ ¹ · ¨ ¨ © § D E D E log 5 log 10 log 10 2 0 jm m R r m j j (13)
Similarly,'T_ADDj can be obtained by Eq. (14).
¸ ¹ · ¨ © § D E Z 'T_ADDj 5 jmlog (14)
4. Simulation experiments
This section analyzes and compares the performance measurements obtained by simulation experiments for CSH, VTSH, ESH and CTSA algorithms, such as system throughput, handoff call dropping ratio, and
new call blocking ratio. We assume that two adjacent base stations are the hot-spot cells having higher mobile density, while the other surrounding cells, eight BSs, are the general cells having lower average mobile density. Table 1 shows the related parameters for system shedding and recovery used in the simulation.
Table 1 Parameters for shedding and recovery
Parameters Values
Base station service radius 1000 m Base station transmission bandwidth 9 Mbps Call arrival process Poisson Guard channel for soft-handoff call 2%
Į 0.95 ȕ 0.9 Ȗ 0.85 Traffic load shedding parameters Ȧ 0.5
Figure 5 shows system throughput against call arrival rate according to the sending or receiving data by a mobile. The hot-spot situation begins at the traffic load 500 calls/min and CTSA performs best, but CSH is worst. CSH has the most quantity of handoff calls and the least system throughput, because it does not have shedding function. VTSH cannot effectively keep shedding traffic when traffic load increases such that its improvement is restricted, because the handoff region is adjusted only once when traffic load is getting heavier. ESH dynamically begins adjusting the handoff region at the traffic load over 70% of the hot-spot traffic (i.e., 350 calls/min); it cannot adjusts the service region to the optimal status after having the hot-spot situation, because ESH occasionally adjust the handoff region according to the traffic load variation. When the hot-spot situation occurs, CTSA first adjust handoff threshold parameters and then adjust handoff region after waiting for a period depending on the traffic load
Figure 5 System throughput of user’s viewpoint against call arrival rate
Figure 6 System soft-handoff call dropping ratio against call arrival rate
Figure 7 System new call blocking ratio against call arrival rate
situation; CTSA also starts postponing algorithm that the mobile located in the neighboring cells postpones to handoff to the hot-spot cell. In summary, CTSA can effectively alleviate BS’s congestion problem.
Figures 6 and 7 show soft-handoff call dropping ratio and new call blocking ratio against call arrival rate, respectively. Obviously, CTSA can effectively execute shedding algorithm and postpones the mobile to handoff to the hot-spot cell according to traffic load variation. CTSA can efficiently reduce the number of handoff calls and soft-handoff call dropping ratio, hence resources can be assigned to new calls to reduce new call blocking ratio. This system reserves 2% of total bandwidth for soft-handoff call so that handoff call has higher priority over new call; therefore the new call blocking ratio is greater than soft-handoff call dropping ratio. CSH performs the worst for both handoff call dropping ratio and new call blocking ratio when traffic load becomes heavier, because CSH
cannot dynamically adjust handoff threshold parameters according to the traffic load. Bandwidth is doubly used in CSH algorithm, which is getting more serious especially for the lack of resource when the number of soft-handoff call requests increases; therefore the handoff call dropping ratio and new call blocking ratio obviously grow up. VTSH performs worse than ESH, because ESH can adjust the handoff region many times according to the traffic load variety but VTSH adjusts only once.
In summary, CTSA can effectively alleviate BS’s congestion problem according to the previous results. It increases system throughput and reduces soft-handoff call dropping ratio, and new call blocking ratio to resolve the hot spot problems.
5. Conclusion
This paper mainly focuses on solving the unbalanced traffic problems for mobile communication system when the number of mobiles rapidly increases in a specific cell. The handoff process of traditional CDMA may make mobiles suffer from serious call blocking and dropping, because BS has no enough resource for the rapid increasing traffic load. CTSA can improve the unbalanced traffic problems by adjusting handoff threshold dynamically according to traffic load variation.
Generally, adjustment range should be moderate; otherwise the service hole problem may be rather serious if it is too large, or the hot-spot traffic cannot effectively be shed if it is too small. System can save additional capacity, which may be occupied during the soft handoff process, to reduce call refusing ratio if the mobiles located in the hot-spot cell can handoff to the neighboring cells earlier. However, ping-pong effect may occur and affect system efficiency if the handoff threshold parameters are adjusted too often. CTSA provides shedding, recovery, and postponing algorithms to prevent the ping-pong effect.
References
[1] V. K. Garg, “Wireless Network Evolution 2G to 3G,” Prentice Hall PTR, pp. 219-256, 2002. [2] W. I. Kim, and C. S. Kang, “A new traffic-load
shedding scheme in the WCDMA mobile communication systems,” IEEE Vehicular Technology Conference, vol. 4, pp. 2405–2409, Sept. 2002.
[3] Y. Argyropoulos, S. Jordan, and S. P. R. Kumar, “Dynamic channel allocation in interference-limited cellular systems with uneven
traffic distribution,” IEEE Trans. on Vehicular Technology, vol. 48, no. 1, pp. 224–232, Jan. 1999. [4] X. H. Chen, “Adaptive traffic-load shedding and its capacity gain in CDMA cellular systems,” IEE Communications, vol. 142, no. 3, pp. 186–192, June 1995.
[5] W. Park, H. Um, J. Ahn, and S. Lee, “Performance analysis on traffic load shedding schemes for mobile communication system,” IEEE Personal Communications, vol. 1, pp. 306–310, Oct. 1997. [6] W. I. Kim, and C. S. Kang, “An adaptive soft
handover algorithm for traffic-load shedding in the WCDMA mobile communication system,” IEEE Wireless Communications and Networking, vol. 2, pp. 1213–1217, March 2003.
[7] S. H. Hwang, S. L. Kim, H. S. Oh, C. E. Kang, and J. Y. Son, “Soft handoff algorithm with variable thresholds in CDMA cellular systems,” IEE Electronics Letters, vol. 33, no. 19, pp. 1602–1603, Sept. 1997.
[8] X. Yang, S. Ghaheri-Niri, and R. Tafazolli, “Enhanced soft handover algorithms for UMTS system,” IEEE Vehicular Technology, vol. 4, pp. 1539–1543, Sept. 2000.
[9] B. Natarajan, C. R. Nassar, S. Shattil, Z. Wu, and M. Michelini, “High performance MC-CDMA via carrier interferometry codes,” IEEE Trans. on Vehicular Technology, vol. 50, no. 6, pp. 1344–1353, Nov. 2001.
[10] Seyed Alireza Zekavat, Carl R. Nassar, and Steve Shattil, “Merging Multicarrier CDMA and Oscillating-Beam Smart Antenna Arrays: Exploiting Directionality, Transmit Diversity, and Frequency Diversity,” IEEE Trans. on Comm., vol. 52, no. 1, pp. 110–119, Jan. 2004.
[11] C. S. Kang, K. S. Kim, and M. J. Kim, “Method For Shedding Traffic Load in Code Division Multiple Access Mobile Communication System,” US Patent 5-781-861, July 1998.
[12] H.G. Jeon, S. H. Hwang, S. K. Kwon, and C. E. Kang, “A channel assignment scheme for reducing call blocking rate in a DS-CDMA cellular system,” IEEE Personal Communications, vol. 2, pp. 637–641, Oct. 1997
[13] Andrew J. Viterbi, CDMA: Principles of Spread Spectrum Communication, Addison Wesley, pp. 179-233, 1995.
