Performance Evaluation and Discussions

Fig 4.1 is the BLER versus HSPR with fixed user mobility at 60 km/hr. The motion incurs the Doppler Effect and the channel variance with CQI delay. Hence, the actual channel condition will be different from the channel information used for determination. It can be seen in Fig 4.1 that the proposed FQLM-HARQ satisfies the BLER requirement with HSPR more than 60%, the next is QL-HARQ with HSPR more than 70%. However, the FTS violates the BLER requirement even with HSPR up to 80%. The FQLM-HARQ has better learning way by considering the situation in different part of BLER n and then adjusts the selection of MCS ( )

40

level to do its possible maintaining the BLER requirement. As the more BS transmission power is allocated for HS-DSCH service, the more MCS level can adaptively select in Red Part to reduce the effect of the channel variance. The MCS level selection of FTS just depends on the current channel condition while it is regardless of the channel variation. This results the BLER performance violating the BLER requirement. Although QL-HARQ also adjusts the MCS level based on the last transmission decision, it does not take the information of transmission results into account. Therefore, it is not flexible enough to accommodate to the channel variation like FQLM-HARQ.

Fig 4.2 is the dropping rate versus HSPR with fixed user mobility at 60 km/hr. In the simulation, when the total transmission times including retransmissions of the same transmission block is more than three, this block will be dropped. It can see that FQLM-HARQ has the lowest dropping rate and FTS has the highest dropping rate despite the HSPR. This has the relation between initial BLER shown in Fig 4.1 and the MCS level selection with the conservative or aggressive way. The smaller BLER performance can result in the lower dropping rate. In FQLM-HARQ, once the dropping block occurs, the design of the reinforcement signal will give the most punishment at each part according the value of

( )

BLER n . After updating the Q-function, we can expect conservative MCS level selection. This design can limit the dropping rate. The QL-HARQ just considers the difference between the received SINR and the required SINR at the last transmission. Therefore, the MCS level will be more aggressive than the FQLM-HARQ and results the more dropping rate.

Fig 4.3 is the system throughput versus HSPR with fixed user mobility at 60 km/hr. It can see that when HSPR increases, the throughputs of the three schemes increase, absolutely. With Fig 4.1, Fig 4.2 and Fig 4.3, FQLM-HARQ can select the optimal MCS level for the largest throughput than the other two schemes while endeavoring to maintain the BLER requirement and result the least dropping rate simultaneously.

41

Figure 4.1: The BLER versus HSPR with fixed user mobility at 60 km/hr.

Figure 4.2: The dropping rate versus HSPR with fixed user mobility at 60 km/hr.

42

Figure 4.3: The system throughput versus HSPR with fixed user mobility at 60 km/hr.

In the following, we will discuss about the different user mobility with fixed HSPR. Fig 4.4 shows the BLER versus HSPR for FTS with different user mobility. There are two characters in this figure. As the user mobility increasing, the BLER will increase with the same HSPR. As the HSPR increasing, the difference of BLER between the different mobility will increase. When HSPR is more than 70%, the BLER of the same user mobility will not increase. This is due to the power achieve the saturation despite the distance between the user and the BS. On the other hand, the total number of MCS levels used in our system is 8 and the maximum MCS level will not reflect the channel quality. This limits the BLER performance. Therefore, we consider the effect of different user mobility with fixed HSPR at 60%.

43

Figure 4.4: The BLER versus HSPR for FTS with different user mobility.

Fig 4.5 is the BLER versus different user mobility with fixed HSPR at 60%. When the user is motionless, FQLM-HARQ has the largest BLER than the other two schemes while not violating the BLER requirement. Because the channel quality is very similar between two adjacent TTI, the design of FQLM-HARQ will try to transmit the more aggressive MCS level that the channel condition can tolerate without dropping. When the user mobility is from 0 km/hr to 30 km/hr, the BLER of three schemes rush up owing to the path loss effect. The increasing range of FQLM-HARQ is the least and this means it can learning much well than the other two schemes. After 30 km/hr, the BLER of three schemes will increase due to the increasing channel variation.

Fig 4.6 is the dropping rate versus different user mobility with fixed HSPR at 60%. The increasing mobility will induce the larger dropping rate for three schemes. As the same reason, the higher channel variation will results unexpected channel condition at next TTI. Therefore, even the MCS level determination from learning will have higher probability to be dropped.

Fig 4.7 is the system throughput versus different user mobility with fixed HSPR at 60%. As the mobility during 0 km/hr to 30 km/hr, the purpose of MCS level selection in this region is to balance the BLER and throughput. When the mobility is more than 30 km/hr, the FQLM-HARQ

44

is trying to reach the QoS requirement and will select the more conservative MCS level. After 90 km/hr, the throughput of three schemes almost not increase due to the learning speed can’t catch up the channel variation.

Figure 4.5: The BLER versus different user mobility with fixed HSPR at 60%.

Figure 4.6: The dropping rate versus different user mobility with fixed HSPR at 60%.

45

Figure 4.7: The system throughput versus different user mobility with fixed HSPR at 60%.

46

5555

Chapter 5 Conclusion

Fuzzy Q-learning based MIMO HARQ scheme (FQLM-HARQ) is proposed to achieve efficient resource utilization in MIMO HSDPA system. The hybrid ARQ can be modeled as a Discrete-time Markov decision process (MDP). We apply fuzzy Q-learning algorithm to adjust the MIMO transmission mode and MCS level selection of initial transmission each TTI. The fuzzy rule is designed based on the channel quality indicator and BLER performance. The reinforcement signal is designed not only according to BLER performance but also considering the past transmission results. These can much help to accommodate the channel variation and the channel quality delay. By self-learning step by step, FQLM-HARQ can expect to maximize the system throughput while not violating the BLER requirement by choosing the optimal MCS level for each transmission.

From simulation results, with fixed user mobility, FQLM can have better performance than the other two schemes in different HSPR. Because of the BLER performance will affect the fuzzy rule base and then considering the different MCS level selection. When BLER is higher, the MCS level selection policy will become more conservative to satisfy the BLER requirement.

On the other hand, when BLER is lower, the MCS level selection policy will become more aggressive to increase the system throughput. As the same reason, while in different user mobility with fixed HSPR, FQLM-HARQ also has more ability to resist the imprecise channel

47

quality. Finally, the analysis shows that FQLM-HARQ scheme can achieve our designed target.

48

Bibliography

[1] 3GPP TS 25.214, “UMTS Physical layer procedures (FDD)” 3rd Generation Partnership Project, Tech. Rep., May 2008.

[2] D. G. Brennan, “Linear diversity combining techniques,” Proc. IRE, vol. 91, no. 2, pp.

331-356, Feb. 2003.

[3] S. Parkvall and E. D. Hlman, “Performance comparison of HARQ with chase combining and incremental redundancy for HSDPA,” Proc. IEEE VTC-Spring 2001, pp. 1829-1833.

[4] T. Cheng, “Coding performance of hybrid ARQ schemes,” IEEE Trans. Commun., vol. 54, no. 6, pp. 1017-1029, Jun. 2006.

[5] M. Peng and W. Wang, “Advanced HARQ and scheduler schemes in TDD-CDMA HSDPA systems,” Proc. the 10th Asia-Pacific Conf. Commun. and 5th Int. Symposium on Multi-Dimensional Mobile Commun., vol. 1, pp. 67-70, Aug. 2004.

[6]. M. Nakamura, Y. Awad, and S. Vadgama, “Adaptive control of link adaptation for high speed downlink packet access in W-CDMA,” Wireless Personal Multimedia Commun., vol. 2, pp. 382-386, 2002.

[7] A. Muller and T. Chen, “Improving HSDPA link adaptation by considering the age of channel quality feedback information,” Proc. IEEE VTC-Fall 2005, pp. 1643-1647.

[8] C. J. C. H. Watkins and P. Dayan, “Q-learning,” Machine Learning, vol. 8, pp. 279-292, 1992.

[9] C. J. Chang, C. Y. Chang, and F. C. Ren, “Q-learning-based hybrid ARQ for high speed downlink packet access in UMTS,” Proc. of VTC-Spring 2007, pp.2610-2615.

[10] G. J. Foschini, “Layered space-time architecture for wireless communication in fading

49

environment when using multiple antennas,” Bell Labs Syst. Tech. J., vol. 1, no. 2, pp. 41-59, Autumn 1996.

[11] V. Tarokh, H. Jafarkhani, and A. R. Calderbank, “Space-time block codes from orthogonal designs,” IEEE Trans. Inform. Theory, vol. 45, no. 5, pp. 1456-1467, Jul. 1999.

[12] L. Zheng and D. N. C. Tse, “Diversity and multiplexing: a fundamental trade off in multiple-antenna channels,” IEEE Trans. Inform. Theory, vol. 49, no. 5, pp. 1073-1096, May 2003.

[13] R. T. Derryberry, S. D. Gray, D. M. Ionescu, G. Mandyam, and B. Raghothaman,

“Transmit diversity in 3G CDMA systems,” IEEE Commun. Magazine, pp. 68-75, Apr. 2002.

[14] J. Peisa, S. Wager, M. Sågfors, J. Torsner, B. Göransson, T. Fulghum, C. Cozzo, and S.

Grant, “High speed packet access evolution – concept and technologies,” Proc. IEEE 65th VTC-Spring 2007, pp. 819-824.

[15] H. Holma, A. Toskala, K. Ranta-aho, and J. Pirskanen, “High-speed packet access evolution in 3GPP release 7,” IEEE Commun. Magazine, pp. 29-35, Dec. 2007.

[16] T. Nihtil¨ and V. Haikola, “HSDPA MIMO system performance in macro cell network,”

IEEE Sarnoff Symposium, 2008.

[17] S. Semenov, “Modification of the D-TxAA scheme for fading channel,” Proc. of the Fourth Advanced Int. Conf. on Telecommun., 2008.

[18] E. N. Onggosanusi, A. G. Dabak, Y. Hui, and G. Jeong, “Hybrid ARQ transmission and combining for MIMO systems,” Proc. IEEE Int. Conf. Commun., vol. 5, pp. 3205-3209, May 2003.

[19] F. Wang, T. Peng, and W. Wang, “Multiple-input multiple-output system antenna subset selection with HARQ,” Proc. IEEE 18th Int. Symposium on PIMRC, Sep. 2007.

[20] P. Y. Glorennec and L. Jouffe, “Fuzzy Q-Learning,” Proc. IEEE Int. Conf. Commun. on Fuzzy System, vol. 2, pp. 659-662, 1997.

50

[21] Y. Maeda, “Fuzzy adaptive Q-Learning method with dynamic learning parameter,” Proc. of IFSA World Congress and 20th NAFIPS International Conf., vol. 5, pp.2778-2780, 2001.

[22] M. Fei and S. L. Ho, “Progress in on-line adaptive, learning and evolutionary strategies for Fuzzy logic control,” IEEE Int. Conf. on Power Electonics and Drive Systems, vol. 2, pp.1108-1113, 1999.

[23] C. J. Chang, C. Y. Chang, and F. C. Ren, “Q-learning-based hybrid ARQ for high speed downlink packet access in UMTS,” Proc. of VTC-Spring 2007, pp. 2610-2615.

[24] L. Gordon, “Principle of Mobile Communication”, Kluwer Academic publishers, 2001.

[25] M. Gudmundson, “Correlation model for shadow fading in mobile radio systems,” IEEE Electron. Lett., vol. 27, pp. 2145-2146, Nov. 1991.

[26] B. Josef and F. Ivan, “CQI reporting for MIMO transmission in a wireless communication system” U.S. WO 6374 July 17, 2008.

51

Vita

Ying-Yu Chen was born in 1983 in Chunghwa, Taiwan. She received the B.E. degree in electrical and engineering from National Chung Hsing University, Taichung, Taiwan, in 2007 and M.E. degree in communication engineering from National Chiao Tung University, Hsinchu, Taiwan, in 2009, respectively. Her research interests include radio resource management and wireless communication systems.