Chapter 5 Simulation Results and Discussions
5.4 Simulation Results
Suppose that one call request can only connect to one access network at a time here. For each cell, assume the new call arrival rate of conversational, streaming, interactive, and background traffic class calls in the heterogeneous network are
, , , and
1/ 40
AR× AR×1/120 AR×1/120 AR×1/ 240 (users/second), respectively, where AR is the equivalent arrival rate. In the simulation, AR is chosen from 1, 3, 5, 7, and 9.
Fig. 5.1 shows the new call blocking rate. It can be found that UGT has lower new call blocking rate. That is because UGT chooses lower traffic load network with higher probability than iterative TOPSIS in order to achieve load balance. Iterative TOPSIS also takes the loading intensity (utilization) into consideration, but the final decision is influenced by other attributes. The result shows UGT has a little better performance in the new call blocking rate, generally. However, in the high traffic load, the performance is almost the same.
The handoff call blocking rate is illustrated in Fig. 5.2. It seems that UGT has higher handoff blocking rate than iterative TOPSIS. However, Fig. 5.3 (a) and (b), which depict the number of total handoff calls and the number of failed handoff calls, respectively, show that UGT not only has fewer total handoff calls, but also fewer failed handoff calls. This means UGT has lower number of forced terminated calls. So, in fact, UGT is not worse than iterative TOPSIS.
1 2 3 4 5 6 7 8 9 0
0.05 0.1 0.15 0.2 0.25 0.3
Equivalent arrival rate (AR)
New call blocking rate
UGT
iterative TOPSIS
Fig. 5.1 : New call blocking rate
1 2 3 4 5 6 7 8 9
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Equivalent arrival rate (AR)
Handoff call blocking rate
UGT
iterative TOPSIS
Fig. 5.2 : Handoff call blocking rate
1 2 3 4 5 6 7 8 9
Number of total handoff calls
UGT
iterative TOPSIS
(a) Number of total handoff calls
1 2 3 4 5 6 7 8 9
Number of failed handoff calls
UGT
iterative TOPSIS
(b) Number of failed handoff calls
Fig. 5.3 : (a) Number of total handoff calls (b) Number of failed handoff calls
Moreover, it can be found that the trends of new call blocking rate and handoff call blocking rate are very different. That is because the system always reserves 5%
resource for handoff calls. When the normalized loading intensity of one network exceeds 95%, a new call will be blocked immediately. On the contrary, a handoff call will not be blocked until the normalized loading intensity reaches 100%. This causes the new call blocking rate will rise exponentially, but handoff call blocking rate will close to the saturated line when the arrival rate gets high gradually.
Handoff occurrence frequency, defined as the number of handoffs per call, is shown at Fig. 5.4. Generally, UGT has lower handoff occurrence frequency than iterative TOPSIS. The result comes from that UGT takes the mobility into consideration, iterative TOPSIS does not. It can be found that the non-real time call has higher handoff frequency in UGT than that in iterative TOPSIS. Since the real time call is more sensitive to the occurrence of handoff, UGT will do it best to avoid handoff for real time call. That is the penalty weight has higher influence for real time call than non-real time call in UGT.
w
Fig. 5.4 : Handoff occurrence frequency
Fig. 5.5 shows the total throughput and throughput of each network. It can be found that iterative TOPSIS has higher throughput than UGT, and the main difference comes from the throughput in WCDMA and WLAN. The phenomenon can be explained by observing Fig. 5.6 (a) and (b), which plot the number of calls and the number of non-real time calls, respectively. First, the number of calls in WCDMA is analyzed. It can be found that iterative TOPSIS has fewer calls in WCDMA.
Moreover, they are almost non-real time calls. On the contrary, UGT has more number of calls in WCDMA, and they are almost real time calls. In the low traffic load, the allowed data rate exceeds the calls’ requirement a lot in WCDMA. Since iterative TOPSIS has more non-real time calls in WCDMA and the FTP calls always come with burst, the throughput will get higher obviously. When it comes to the calls in WLAN, it can be found there are more WLAN calls for iterative TOPSIS than that for UGT. Since the number of calls has not achieved its capacity, iterative TOPSIS will have higher throughput clearly.
1 2 3 4 5 6 7 8 9
Fig. 5.5 : Total throughput and throughput of each network
1 2 3 4 5 6 7 8 9
Number of non-real time calls
UGT : WCDMA
(b) Number of non-real time calls
Fig. 5.6 : (a) Number of calls (b) Number of non-real time calls
The average delay for voice and video call in the heterogeneous network are shown in Fig 5.7 and 5.8, respectively. It can be found that the average delay for voice call is almost the same. That is because this traffic class call is highest priority.
However, the average delay for video call in high traffic load is higher for UGT than that for iterative TOPSIS in WMAN. This is because there are more video calls for UGT than that for iterative TOPSIS in high traffic load in WMAN. In this situation, WMAN may not have enough resource when a burst comes for video streaming, and then their delay will get high.
1 2 3 4 5 6 7 8 9
0 5 10 15 20 25 30 35 40
Equivalent arrival rate (AR)
Average delay of voice traffic (ms)
Delay requirement UGT-WCDMA UGT-WMAN UGT-WLAN
iterative TOPSIS-WCDMA iterative TOPSIS-WMAN iterative TOPSIS-WLAN
Fig. 5.7 : Average delay of voice traffic
1 2 3 4 5 6 7 8 9
Average delay of video traffic (ms)
Delay requirement
Fig. 5.8 : Average delay of video traffic
The average dropping rate for voice and video call are shown in Fig. 5.9 and Fig.
5.10, respectively. It can be found that the maximum packet dropping rate requirement is satisfied for each scheme. However, the dropping rate is higher in UGT than that in iterative TOPSIS. This is because UGT sees those networks as the same if they can provide enough good QoS requirement, just as shown in Fig. 4.1. On the contrary, iterative TOPSIS see the network as the best if it can provide best QoS requirement for it. Moreover, it can be found that UGT has fewer number of calls in WLAN than iterative TOPSIS has, but the dropping rate is higher in UGT. This is because the calls are almost non-real time calls in UGT. In the design of WLAN in this thesis, it is assumed that when a FTP call gets the right of channel usage, it will transmit 3000 bytes. That is it will occupy at least 12 ms! On the contrary, real time calls transmit much fewer bits than non-real time calls. So the system with more non-real time calls will have higher delay variance. This situation will cause higher dropping rate.
1 2 3 4 5 6 7 8 9
Average dropping rate of voice traffic (%)
Maximum acceptable dropping rate
Fig. 5.9 : Average dropping rate of voice traffic
1 2 3 4 5 6 7 8 9
Average dropping rate of video traffic (%)
Maximum acceptable dropping rate
Fig. 5.10 : Average dropping rate of video traffic
Chapter 6
Conclusions and Future Works
In this thesis, a utility and game-theory (UGT) based network selection scheme is proposed for heterogeneous wireless access network. By considering four multimedia services, including conversational, streaming, interactive, and background, a call admission control is performed first to find which network can be used when a call request comes. After getting the set of candidate networks, a utility value is obtained to represent the satisfaction degree of QoS requirement. Moreover, in order to achieve load balance and consider mobility factor, a cooperative game is defined to get the preference value for each network. Finally, the most suitable network for the call request can be decided by linear combination of above set of values.
Simulation results show that UGT has lower total throughput than iterative TOPSIS while satisfying the QoS requirements of each traffic class. As known, the difference mainly comes from the non-real time calls. By sacrificing little throughput of non-real time calls, UGT can obtain lower new call blocking rate, fewer forced terminated calls, and fewer handoff occurrence frequency. Lower new call blocking rate and fewer forced terminated calls mean that the heterogeneous system can accommodate more calls. Besides, UGT reduces the handoff occurrence frequency about 30% than iterative TOPSIS generally. However, this value even exceeds 50%
for real time calls! With lower handoff occurrence frequency, some problems,
happening during the processing of handoff calls, can be avoided substantially. In this aspect, iterative TOPSIS is overwhelmed by UGT. When it comes to the dropping rate, UGT is higher than iterative TOPSIS obviously. But they are all under the maximum acceptable dropping rate. Allowing a little higher dropping rate to exchange for other better performance, which is more critical, may be very worthful. The interesting phenomenon can be observed in our simulation results.
The work can be extended to vertical handoff problem. In this thesis, it is assumed that the handoff occurs only when the call is out of the coverage the original network. However, the handoff can be performed in advance to get better system performance, just like [7]. To make the handoff decision, UGT can be used. At each observation period, an existing call must to decide whether it needs to hand off or not.
Some modification of UGT may be very suitable for this problem.
Appendix A
KKT Conditions
Consider the following maximizing problem which subjects to equality and inequality constraints:
Note that and is the set of positive integers. The Karush-Kuhn-Tucker (KKT) condition [24] can be used to find the solution of above problem. Define λ as the Lagrange multiplier vector, and μ∈Rn as the KKT multiplier vector. The KKT condition consists of five parts (three equality and two inequality equations), and is given below
1) μ 0≥ ,
2) Df(NP)+λ⋅Dh(NP)+μTDg(NP)= , where D is the derivative operator. 0 3) μ g NPT ( )=0 ,
4) h(NP)=0 , 5) g NP( )≥0,
Put (A.1) into the KKT condition, then the following results are obtained three cases to solve the problem.
Case 1: No value of μ is equal to 0.
That is μi > ,0 for i=1 ~n, and NPi =0, for i=1 ~n. The result conflicts with (A.5), so the set of solution is impossible.
Case 2: Only one value of μ is equal to 0.
Assume 0μi = , where i∈{1, 2, , }" n ; 0μj > , NPj =0, for j=1 ~ , n j≠ . i From (A.5), NPi = . From (A.3), 1 λ=Ai(2wi− , ,1) μj = − −λ A for jj =1 ~ ,n j≠ . i The values of μ must be checked that whether they satisfy (A.2) or not. If satisfied, then this set of solution is valid.
Case 3: More than one value of μ is equal to 0.
Put the results into (A.3), the following equations are obtained:
(A.7) respectively. If satisfied, then this set of solution is valid.
NP
In fact, there are total (2n− situations (excluding the situation which all NP 1) equal to 0). For each situation, check whether the solution satisfies the KKT condition or not. Because this function is a quadratic equation, the solution which satisfies the KKT condition must be the optimal solution.
Bibliography
[1] E. Stevens-Navarro, Y. Lin, and V. W. S. Wong, “An MDP-based vertical handoff decision algorithm for heterogeneous wireless networks,” IEEE Transactions on Vehicular Technology, Vol. 57, no. 2, pp. 1243-1254, 2007.
[2] W. Song, H. Jiang, W. Zhuang, and X. Shen, “Resource management for QoS support in cellular/WLAN interworking,” IEEE Network, Vol. 19, no. 5, pp. 12-18, Sept.-Oct. 2005.
[3] Y. H. Chen, N. Y. Yang, C. J. Chang, and F. C. Ren, “A utility function-based access selection method for heterogeneous WCDMA and WLAN networks,”
IEEE 18th International Symposium on PIMRC-Sept. 2007, pp. 1-5.
[4] F. Bari and V. C. M. Leung, “Automated network selection in a heterogeneous wireless network environment,” IEEE Network, Vol. 21, no. 1, pp. 34-40, Jan.-Feb.
2007.
[5] O. Yilmaz, A. Furuskar, J. Pettersson, and A. Simonsson, “Access selection in WCDMA and WLAN multi-access networks,” IEEE VTC-Spring 2005, Vol. 4, pp.
2220-2224.
[6] J. Chen, K. Yu, G. Yang, Z. Y. Feng, Y. Ji, P. Zhang, Q. Huang, Y. Bai, L. Chen, and M. Minomo, “An ecology-based adaptive network control scheme for radio resource management in heterogeneous wireless networks,” BIMNICS-Dec. 2006, pp. 1-5.
[7] G. Ning, G. Zhu, L. Peng, and X. Lu, “Load balancing based on traffic selection in heterogeneous overlapping cellular networks,” The First IEEE and IFIP International Conference in CANET, Sept. 2005.
[8] G. Ning, G. Zhu, Q. Li, and R. Wu, “Dynamic load balancing based on sojourn time in multitier cellular systems,” IEEE VTC-Spring 2006, Vol. 1, pp. 111-116.
[9] A. Dixit and S. Skeath, “Games of strategy,” New York:W.W. Norton, 2004.
[10] J. Antoniou and A. Pitsillides, “4G Converged Environment: Modeling Network Selection as a Game,” IEEE Proceeding on IST Conf., pp. 1-5, July 2007.
[11] D. Niyato and E. Hossain, “A cooperative game framework for bandwidth allocation in 4G heterogeneous wireless networks,” IEEE ICC-June 2006, Vol. 9, pp. 4357-4362.
[12] M. Pulido, J. S. Soriano, and N. Llorca, “Game theory techniques for university management: An extended bankruptcy model,” Operation Research, Vol. 109, pp.
129-142, 2002.
[13] D. Niyato, and E. Hossain, “A noncooperative game-theoretic framework for radio resource management in 4G heterogeneous wireless access networks,” IEEE Transactions on Mobile Computing, Vol. 7, no. 3, pp. 332-345, March 2008.
[14] 3GPP TR 23.271, “Functional stage 2 description of Location Services (LCS),”
3rd Generation Partnership Project, Tech. Rep., Sep. 2007.
[15] H. Holma and A. Toskala, “WCDMA for UMTS,” John Wiley and Sons, Ltd, 2002.
[16] IEEE Std. 802.16e, “IEEE standard for local and metropolitan area networks part 16: air interface for fixed broadband Wireless access systems amendment for physical and medium access control layers for combined fixed and mobile operation in licensed bands,” Oct. 2005.
[17] A. J. Goldsmith and S. G. Chua, “Variable-rate variable-power MQAM for fading channels,” IEEE Transactions on Communications, Vol. 45, pp.1218-1230, Oct. 1997.
[18] IEEE Std. 802.11e/D12.0, “IEEE Standard for Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications: Medium Access Control (MAC) Enhancements for Quality of Service (QoS),” Nov. 2004.
[19] 3GPP TR 25.892, “Feasibility study for OFDM for UTRAN enhancement,” 3rd
Generation Partnership Project, Tech. Rep., 2004-06.
[20] L. Gordon, “Principle of Mobile Communication,” Kluwer Academic, 1958 [21] 3GPP TR 23.107, “QoS concept and architecture release 6,” 3rd Generation
Partnership Project, Tech. Rep., March 2004.
[22] Universal Mobile Telecommunication System, “Selection procedures for the choice of radio transmission technologies of the UMTS”, UMTS Std. 30.03, 1998.
[23] R. Guerin, H. Ahmadi, and M. Naghshineh, “Equivalent capacity and its application to bandwidth allocation in high-speed networks,” IEEE Journal on selected Areas in Communications, Vol. 9, pp. 968-981, Sep. 1991.
[24] E. K. P. Chong and S. H. Zak, “An introduction to optimization,” John Wiley and Sons, Inc, 2001.
Vita
Tsung-Li Tsai was born in 1984 in Changhua, Taiwan. He received the B.E.
degree in electrical engineering from National Cheng-Kung University, Tainan, Taiwan, in 2006, and the M.E. degree in the department of communication engineering, college of electrical and computer engineering from National Chiao Tung University, Hsinchu, Taiwan, in 2008. His research interests include radio resource management and wireless communication.