Mapping Function from User Perspective

In different networks, the same signal-to-noise ratio (SNR) might have different throughput performances. Therefore a direct comparison of the SNR values will cause a misinterpretation of the resulting performance. To resolve the problem, an effective SNR, based on the same throughput reference, could be used. For example, as shown in Figure 3-1, based on a WLAN performance curve, a mapping between the achievable rates from UMTS and WLAN, RUMTS and RWLAN, and the corresponding effective SNR values, SUMTS and SWLAN

can be identified. In short, the original SNR is first used to estimate the achievable throughput which depends on the RF condition only. Through the mapping module, the corresponding effective SNR, SWLAN and SUMTS, are calculated based on a WLAN throughput performance curve. A dynamic vertical handover control algorithm can then be derived based on the effective SNR. Basically, the vertical handover is triggered when the differences between the S_WLAN and S_UMTS values exceed or drop below a dynamic threshold, H, for a period of time. As stated in Eq. (3-1), the downlink handover from UMTS to WLAN is triggered when the effective SNR from WLAN is greater than that of the UMTS by H for ∆T_downlink seconds.

From Eq. (3-2), for the WLAN to UMTS handover, the uplink handover is triggered when the

effective SNR from the UMTS is greater than that from the WLAN by H for ∆T_uplink seconds, where the threshold H and the trigger timer ∆T_{downlink/uplink} could be a function of QoS requirements and would be addressed in later section.

downlink downlink

UMTS

WLAN S H for T

S − > ∆ (3-1)

uplink uplink

UMTS

WLAN S H for T

S − < ∆ (3-2)

Figure 3-1. Throughput vs SNR in WLAN (including link adaptation)

3.2.2 The General form of the mapping function from system perspective

The throughput performance used in the section 3.2.1 is estimated by the individual measurements of signal strength (RF only). However the actual throughput differs a lot, especially in WLAN system. This is because the medium in WLAN is shared by many users;

Effective SNR Throughput

R^UMTS R^WLAN

S^UMTS S^WLAN S^UMTS S^WLAN R^WLAN

R^UMTS

(1)

(2)

collision and simultaneous backoff stages will occupy the medium and reduce the effective throughput. Furthermore, the throughput in WLAN will be saturated when the loading is high.

Thus, call admission control in WLAN is crucial to maintain the throughput and other QoS performances. In this section, the effect throughput, RE, calculating from the system perspective is used as the general form of the mapping function. In calculating effective throughput from the system perspective, it takes the transmission ratio and call admission control into considerations, the relationship between R and RE is:

R medium). Therefore the transmission probability Pt can be calculated in Eq (3-4)

c

Where P_s is the successful transmission probability with associated successful transmission time T_s, P_c is the collision probability with collision time T_c and E[I] is the average idle period.

The expressions of T_s andT_c in Eq. (3-5) and (3-6) are according to the Figure 3-2 diagram which applied the DCF scheme and RTS/CTS mechanism.

Figure 3-2. The expression of successful transmission and collision time

DIFS ACK

SIFS Payload

Header SIFS

CTS SIFS RTS

T_s = + + + + + + + + (3-5)

Timeout CTS

SIFS RTS

T_c = + + _ (3-6)

The remained work is to get the probabilities for all periods. Here a three dimension Markov-chain model [36-38] based on the number of users, the back-off window stages and process time is used to calculate the probabilities. Finally, substituting all parameters into Eq (3-4), the transmission ratio probability Pt can be calculated.

For the allowable incoming probability, Pa, it depends on the call admission control (CAC) in WLAN system. The main objective of CAC is to prevent channel overload and protect existing users. Therefore the allowable incoming probability, P_a, would be a conditional probability which depends on the new collision rate and the limits set for all existing users. If the new incoming user will cause the average collision rate exceed a

threshold, then CAC will block the incoming flow. To trace the new collision rate, a counter rate before admitting the new incoming user and α is the weighted factor which is designed by users’ experience (usually be 0.8). With this counter, the allowable incoming probability Pa

could be expressed as:

mapping flow is shown in Figure 3-3.

Figure 3-3. The mapping flow

3.3 QoS-Based Dynamic Handover Threshold

To reflect the benefit of the vertical handover in various aspects, besides the effective SNR values, the algorithm needs to consider dynamic thresholds and the associated timers to achieve the QoS requirements. In the proposed vertical handover algorithm, the dynamic threshold, H, will depend on service types of either non-real time services or real-time

services. For non-real-time services, the transmission packets come at a burst and are not sensitive to the delay. In this case, the user transmission rate becomes the first priority for the vertical handover. On the other hand, for real-time services, the services have a stringent requirement in the delay bound. Besides, for WLAN-to-UMTS uplink handover, since WLAN has a smaller coverage, the connectivity becomes essential in the uplink handover.

To achieve above design goals, the dynamic threshold is proposed as:

dB respectively. ∆ is the handover latency (the process delay for a vertical handover). The m, n, and k are values of 0 or 1 which will be decided based on following conditions:



Under this design, for non-real-time services, the throughput ratio provides additional weight on the threshold, H, calculations. This is because the effective SNR in some cases can’t accurately represent the absolutely throughput difference between UMTS and WLAN.

See Figure 3-1, the difference of the effective SNR between UMTS and WLAN represented in blue lines (1) is the same as the red lines (2), but the difference of throughput in red lines is much larger than blue lines’. The reason is that WLAN throughput may reach the saturation point no matter how larger SNR is. So the weighted factor for non-real time services is needed.

The dynamic threshold decreases when the throughput ratio of WLAN and UMTS becomes larger. This will make the UMTS-to-WLAN handover easier. Thus, a mobile with non-real-time services can achieve higher throughput. For real-time services, besides the over-air-error, the packet loss happens when the delay bound expires, the excessive handover delay,∆−D_i, will also cause the loss packets in the real-time services. With the allowable

number of lost packets Bi for application i, the number of lost packets due to the vertical handover, (∆−D )_i T_i (see Figure 3-4) should be less than Bi. If the resulting error

performance exceeds Bi, the effect of the degradation will be considered in the dynamic threshold, H. To emphasize the connection quality, the weighted factor,β , as expressed in Eq _i (3-9) is increased proportionally to the increase of the packet loss.

dB

Figure 3-4. The packet loss for real-time services in handover

Finally, a timer hysteresis,∆T_uplink, is considered to avoid the ping-pong effect in the conventional handover control. However since the coverage in WLAN is small, any excessive delay might result in the discontinuity of the connection. To resolve this potential problem, the dynamic threshold, H, will include (∆+∆T_uplink) ∆ for an earlier uplink handover trigger if the time hysteresis ∆T_uplink is larger. It can be seen that when the time hysteresis is larger, the dynamic handover threshold, H, would be larger. The result forces the early trigger for WLAN-to-UMTS handover, as show in Eq. (3-2).

3.4 Performance-Based Trigger Timer

In conventional handover algorithm, the objective of the trigger timer is to resolve the ping-pong effect [39, 40]. To calculate a proper trigger timer, a performance-based trigger timer, which the length of the trigger timer depends on whether the resulting performance can be improved after the handover, is calculated.

3.4.1 Downlink handover

From Eq. (3-12), the handover from UMTS to WLAN is worthwhile only when the user can transmit more data in WLAN than that in the UMTS after the handover process is finished. To calculate the downlink handover timer,∆T_downlink, it is assumed that the RE WLAN and RE UMTS are stable during this handover period. In this case, the timer∆T_downlinkis calculated from Eq.

The same argument can apply to the uplink handover timer calculation. The uplink handover is worthwhile when Eq. (3-14) is satisfied:

∫

The uplink handover timer,∆T_uplink, is calculated by Eq. (3-15).

UMTS E

WLAN E uplink

R r R

r

T =

−

≥ ∆

∆ ,

1 1

2 (3-15)

The handover timers are not fixed and will be updated based on the value r from time to time.

Figure 3-5. Trigger time functionality for downlink handover

3.5 Summary of Proposed Vertical Handover Control Algorithm

The proposed vertical handover control algorithm is composed by throughput-based mapping function, QoS-based dynamic handover threshold, and performance-based trigger timer. The procedures of the vertical handover control algorithm depicted in Figure 3-6:

1. Measures the signal strength and calculates the effective throughput from system perspective.

Time

∆ ∆

∆T ∆T

Throughput

0 R

^UMTS

R

^WLAN Ping pong effect

(compare throughput performance) handover

No handover

2. Uses the mapping function to get the effective SNR.

3. Applies different services with different QoS requirements to set the dynamic handover threshold and trigger timer for downlink and uplink handover.

4. Finally, trigger equations in Eq (3-1), (3-2) is used to make decision about the handover.

Measurement

Estimation (user perspective)

Calculation (system perspective)

Trigger timer

setting Mapping Threshold setting

Sw - Su > H for ∆Tdownlink Sw–Su < H for ∆Tuplink

services

SNR

R

RE

∆Tuplink,

∆Tdownlink SW, SU H

Yes, Handover

No, stay

Figure 3-6. The procedure of proposed vertical handover algorithm

Chapter 4

Mathematical Model and Numerical Analysis

In this chapter, a mathematical model is created to analyze the handover performance.

Simulation results and the analysis for the proposed vertical handover algorithm are also given

4.1 Mathematical Model

In this section, the performance of the handover frequency and average throughput is analyzed and the impacts from the path loss and shadow fading are considered. The fast fading will be ignored due to the averaging of the measurements. As calculated in Eq. (4-1) and (4-2), the signals (in dB) received at MS from UMTS and WLAN, are U (k) and W (k), respectively.

( ) ( )

dk u dk

K K k

U( )= ₁− ₂log + (4-1)

( )

(

w dk

)

v

(

w

( )

dk

)

K K k

W( )= ₃− ₄log + (4-2)

As shown in Figure 4-1, dk is the distance when the MS is d meters from UMTS at kth sample time and the functionw (dk) calculates the distance from WLAN when the mobile is d meters

from UMTS. K₁, K_2, K_3, and K₄ are parameters for the path loss. The shadow fading, u (d_k) and v (dk) are assumed to be independent and identically distributed stationary Gaussian random processes with zero mean and variance,σ . ²

Figure 4-1. UMTS and WLAN location

The received signal will be averaged by applying an exponential filter, implemented as a low-pass filter:

model with associated probabilities will be used for representing the behavior of the vertical handover. As derived in Eq. (4-5), the state probabilities of UMTS, (PUMTS), WLAN, (PWLAN) and handover probability, (Pho(k)), can be calculated based on both the transition probabilities of UMTS to WLAN, (Pw|u), and WLAN to UMTS, (Pu|w).

Figure 4-2. Handover probability motion

Assuming that MS first connects to UMTS where PUMTS (0) =1, PWLAN (0) =0, it can be seen that Eq. (4-5) can be solved if the probability Pw|u(k) and Pu|w(k) are known.

To solve the transition probability, a vertical handover process is depicted in Figure 4-3.

From Figure 4-3, in UMTS-to-WLAN downlink handover, the transition will occur at the Kth

interval if S_WLAN(K_downlink)−S_UMTS(K_downlink)>H(K_downlink) for ∆T_downlink intervals. Here, if each mapping sample is independent, P_w|u(d_k) can be written as:

)}

Figure 4-3. Finite state machines for handover process

Eq. (4-6) can be also rewritten as

which assumes that the current state and the received strength between the UMTS and WLAN are independent. Similarly, for WLAN-to-UMTS uplink handover, the probability Pw|u(dk) can also be written as: substituting Eq. (4-11) into (4-8) and (4-9), we can then calculate Pw|u(k) and Pu|w(k). Finally, from Eq. (4-5), Pho(k), Pu(k), and Pw(k) can be derived.

When the vertical handover probability Pho(k) and the state probabilities Pu(k) and Pw(k) are calculated, the vertical handover frequency,H , and average throughput, R , for a single f

user can be calculated as follows:

T this model and give discussions.

4.2 Numerical Results and Analysis

With the defined scenario, the performance derived in Eq. (4-5) can then be used to verify the proposed vertical handover control algorithm. Without losing the generality, a hot-spot scenario is assumed where there is only one UMTS base station and one WLAN access point. As shown in Figure 4-1, the trajectory of mobile will across the WLAN coverage.

The system and traffic parameters are listed in Table 4-1:

Table 4-1. Parameters Used for Numerical Analysis

Parameter Value

UMTS radius 500 meter

WLAN radius 50 meter

Separation between UMTS and WLAN 250 meter

K1,K3 (path loss parameter) 0dB

K2,K4 (path loss parameter) 30dB

σ (shadowing variance) 2 6 dB T_av (average distance for filter) 30 ms

∆ (handover latency) 500 ms

Sample time 50 ms

Video traffic delay bound (DB) 50 ms

Video frame per second (1/PI) 25 fps

Allowable frame loss rate (B) 5%

α (parameter in H) 5 dB

In this simulation, both non-real-time services and real-time services are considered. A baseline vertical handover algorithm (based on the signal-strength trigger only) is used as the reference to quantify the proposed vertical handover control algorithm.

4.2.1 Non-real-time services

As discussed, for non-real-time services, achieving higher transmission rates will be the major focus of the proposed vertical handover algorithm. As shown in Figure 4-4, for non-real-time services, the proposed vertical handover control algorithm achieves higher system throughput than the baseline handover control algorithm. As expected from Figure 4-5, the chance of staying in WLAN is higher in the proposed vertical handover control algorithm.

Also, as depicted in Figure 4-6, the proposed vertical handover control algorithm can substantially reduce the handover frequency, which has a positive impact on the processing power and the over-the-air signaling.

0 10 20 30 40 50 60 70 80 90 100 1

2 3 4 5 6 7 8 9

Time (s )

Throughput (Mbps)

proposed

baseline (signal-strength)

Figure 4-4. Throughput statistics for non-real-time services

Figure 4-5. Average state probabilities for different conditions

1 2

Figure 4-6. Performance comparisons for non-real-time services

4.2.2 Real-time services handover frequency. In this case, the proposed algorithm keeps the mobile in the UMTS about 60% of the time when the user initially connected to UMTS, shown in Figure 4-5.

0 10 20 30 40 50 60 70 80 90 100

Figure 4-7. Frame loss statistics (error-free channel) for real-time services

1 2

Figure 4-8. Performance comparisons for real time services

Chapter 5

Conclusions

5.1 Contributions

The challenges in designing the vertical handover are addressed and corresponding solutions are proposed: (1) Throughput-based mapping function is used to resolve the no-common pilot problem in the integrated system. (2) QoS-based dynamic handover thresholds could dynamically change the handover criteria for real-time and non-real-time services. (3) Performance-based trigger timer will calculate the proper trigger timer to avoid excessive ping-pong effect caused by an unstable channel conditions. An analytic model is provided to analyze the handover performance in the heterogeneous networks. Finally, the results show the proposed algorithm could improve the transmission throughput for non-real-time services and could substantially reduce the packet loss rate for the real-time services by reducing the vertical handover frequency.

5.2 Future works

In designing the mapping function, it just considers the throughput performance as a mapping judgment. However the other issues like power budget, building cost, and etc should be also taken into considerations. So a cost function which is a combination of considered parameters with associated weighted factors is expected to be a new judgment for controlling

the vertical handover. Besides, in this scenario, one UMTS cell covers one WLAN cell is assumed. In fact, one UMTS cells would cover several WLAN cells due to the smaller coverage of WLAN system. So how to select the WLAN cell as a handover target is an urgent problem. This issue becomes difficult in uplink handover; this is because mobile can switch to original UMTS network or handover to other WLAN cells. Those two conditions might have different handover methods and performances, so how to choose the most suitable decision for satisfying users’ requirements is a future work. Finally, investigating how to apply this vertical handover control mechanism as a general form for other heterogeneous networks like WLAN/WMAN, Ultra-Wideband/UMTS is a good study in the future works.

Reference

[1] D. J. Goodman “Wireless Personal communications Systems” Addison Wesley Longman 1997.

[2] “BRAN HIPERLAN 3: Requirement and Architectures for Interworking between HIPERLAN/2 and 3^rd Generation Cellular Systems”, ETSI TR 101 957 V1.1.1

[3] G. Corazza, D. Fiancristofaro and F. Santucci, “Characterization of Handover Initialization in Cellular Mobile Radio Networks”, IEEE VTC, 1994

[4] M. Gudumunson, “Analysis of Handover Algorithms”, IEEE VTC, 1991 [5] 3GPP TS 21.103: 3rd Generation mobile system Release 5 specifications [6] 3GPP TS 23.002: Network architecture

[7] 3GPP TS 23.107: Quality of Service (QoS) concept and architecture [8] 3GPP TS 25.301: Radio interface protocol architecture

[9] IEEE Std. 802.11-1999, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications, Reference number ISO/IEC 8802-11:1999(E), IEEE Std. 802.11, 1999 edition, 1999.

[10] IEEE Std 802.11a-1999, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications—Amendment 1: High-speed Physical Layer in the 5 GHz band, Reference number ISO/IEC 8802-11:1999/Amd 1:2000(E)

[11] IEEE 802.11b-1999: Wireless LAN MAC and PHY specifications: Higher speed Physical Layer (PHY) extension in the 2.4 GHz band

[12] IEEE Std 802.11e/D3.0: Wireless Medium Access Control (MAC) and Physical Layer (PHY) specifications: Medium Access Control (MAC) Quality of Service (QoS) Enhancements

[13] H. Bing, C. He and L. Jiang “Performance Analysis of Vertical Handover in a UMTS-WLAN Integrated Networks” IEEE International Symposium on Personal, Indoor and Mobile Radio Communication Proceedings 2003

[14] P. Krishnamurthy, A. Hatami and R. Piclma “Handoff in Hybrid Mobile Data Networks”

IEEE Personal Communications Apr 2000

[15] 3GPP TR 22.934: Feasibility study on 3GPP system to Wireless Local Area Network (WLAN) interworking

[16] M. Gustafsson “Compressed Mode Techniques for Inter-Frequency Measurements in a Wideband DS-CDMA Systems” Proc 8^th IEEE PIMRC, Sep 1997

[17] 3GPP, TR 25.212 V5.5.0: Multiplexing and channel coding (FDD)

[18] 3GPP, TR 25.211 V5.4.0: Physical channels and mapping of transport channels onto physical channels (FDD)

[19] 3GPP, TR 25.215 V5.4.0: Physical layer - Measurements (FDD) [20] 3GPP TSGR1#5(99) 633: User Management in Compressed Mode

[21] United States Patent US6498933, “Method for Transmitting Data for Inter-Frequency/Inter-System Handover in Mobile Telecommunications System”, Dec.

24, 2002

[22] Y. Wang, S. Dan, P. Zhang, H. Wang, “Comparison between the periodic and event-triggered compressed mode”, VTC Spring 2002. IEEE 55th, Volume: 3, 6-9 May 2002

[23] Z. Zhang,” WCDMA compressed mode triggering method for IRAT handover”, WCNC.

2004, IEEE, Volume:2, 21-25 March 2004

[24] IEEE Part 11 “Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications” IEEE Standard 802.11 1999

[25] W. Li, Q. Zeng and P. Agrawal “A Reliable Active Scanning Scheme for the IEEE 802.11 MAC Layer Handoff” Radio and Wireless Conference, 2003. RAWCON '03.

Proceedings

[26] M. Stemm and R. H. Katz “Vertical handoffs in wireless overlay networks” In ACM Mobile Networking (MONET), Special Issu on Mobile Networking in the Internet, 1997 [27] Q. Zhang, C. Guo, Z. Guo, and W. Zhu “Efficient Mobility Management for Vertical

Handoff between WWAN and WLAN” IEEE Communications Magazine November 2003

[28] H. J. Wang, R. H. Katz, and J. Giese “Policy-Enabled Handoffs Across Heterogeneous Wireless Networks” WMCSA, 1999

[29] M. Ylianttila, M. Pande, J. Makela, P. Mahonen “Optimization Scheme for Mobile Users Performing Vertical Handoffs between IEEE 802.11 and GPRS/EDGE networks” Global Telecommunications Conference, 2001

[30] D. B. Johnson. “Scalable support for transparent mobile host internetworking” In Wireless Networks, 1996.

[31] L. Ma, F. Yu, and C.M. Leung “A new method to support UMTS/WLAN vertical handover using SCTP” IEEE Wireless Communications August 2004

[32] K. Kim “Handbook of CDMA Systems Design, Engineering, and Optimization” Prentice Hall PTR 1999

[33] S.W. Wang and I. Wang “Effects of Soft Handoff, Frequency Reuse, and Non-ideal Antenna Sectorization on CDMA System Capacity”, Proc IEEE VTC, May 1993

[34] C.Y. Chung, M.Y. Tsai and J. Huang “Adaptive Quality-based Handoff Algorithms for Third Generation Wireless Communication Systems” IEEE VTC Oct 2004

[35] H.H. Juan and C.Y. Chung “Adaptive Quality-based Handover Control Algorithm in UMTS” IEEE WCE Nov 2004

[36] A. Grilo and M. Nunes, “Performance evaluation of IEEE 802.11e,” in Proc. IEEE PIMRC’02, Sept. 2002, pp. 511-517

[37] S.Choi, J.D. Prado, S. Shankar, and S. Mangold. “IEEE 802.11e contention-based channel access (EDCF) performance evaluation,” in Proc. IEEE ICC’03, vol. 2 May 2003, pp 1151-1156

[38] Z. N. Kong, D. H. K. Tsang, and Brahim Bensaou. ”Performance Analysis of IEEE 802.11e Contention-Based Channel Access,” in Selected Areas in Communications, IEEE Journal on, Volume: 22, Issue: 10, Dec. 2004 pp 2095-2106

[39] S.M. Shirvani, V. Tabataba, and A. Falahati “New handoff initiation algorithm (optimum combination of hysteresis & threshold based methods)” IEEE VTC, 2000

[40] N.C. Chen, R.D. Yates “Evaluation of a minimum power handoff algorithm” Personal, Indoor and Mobile Radio Communications, 1995. PIMRC'95. 'Wireless: Merging onto the Information Superhighway, Sixth IEEE International Symposium on Volume 2, 27-29 Sept. 1995 Page(s):814 - 818 vol.2

在文檔中 針對WLAN與UMTS的新穎異質切換控制演算法 (頁 29-0)