Seamless High-Velocity Handover Support in Mobile WiMAX Networks

Transcription

1 Seamless High-Velocity Handover Support in Mobile WiMAX Networks Zhiwei Yan, Lei Huang and C.-C. Jay Kuo Xi an Jiaotong University, Xi an 749, P. R. China Loyola Marymount University, Los Angeles, CA , USA University of Southern California, Los Angeles, CA , USA s: and Abstract The IEEE 82.6e standard (i.e., mobile WiMAX) has been proposed to provide connectivity in wireless networks for mobile users (including users at a vehicular speed). It is shown in our analysis that the probability of a successful handover decreases significantly when the user moves at a higher speed. Then, we propose a scheme that combines adaptive forward error correction (FEC) with retransmission to offer extra protection for handover signaling messages to enhance the probability of a successful handover, especially at a higher velocity. It is demonstrated by computer simulation that the proposed scheme provides a higher successful handover probability at various velocities. I. INTRODUCTION The rapid growth of high data rate services in wireless communication networks demands new technologies for broadband wireless access. The Worldwide Interoperability for Microwave Access (WiMAX) is an emerging technology that enables the delivery of the last mile wireless broadband access as an alternative to the wired broadband access such as cable and DSL. It provides a convenient way to build a wireless metropolitan area network (WMN), which offers a broad range of high data rate applications, such as broadband Internet access, Voice over Internet Protocol (VoIP), Internet Protocol Television (IPTV), etc. to wireless users anytime and anywhere. For more technical details, we refer to the newly developed IEEE 82.6 standard [], [2]. Handover is one of essential issues in mobile wireless communications since it is needed to maintain uninterrupted services during user s movement from one location to another [3], [4]. It manages mobility between subnets in the same network domain (micro-mobility) and between two different network domains. The current mobile WiMAX standard defines handover operations to support micro-mobility for the point-to-multipoint (PMP) mode communication. That is, it handles the mobile subscriber station (MS) switching from one base station to another. In general, handover techniques can be divided into soft handover (SHO) and hard handover (HHO). SHO employs a make-before-break approach where a connection to the next base station (BS) is established before an MS releases an ongoing connection to the original BS. It guarantees zero break-time during handover at the cost This work is supported in part by the China Scholarship Council (CSC). of lower spectral efficiency. This technique is suitable for latency-sensitive services such as voice communications, video conferencing, and multi-player gaming. In contrast, HHO uses a break-before-make approach where a connection with the serving BS is terminated before a mobile station switches to another BS. HHO is more bandwidth-efficient than SHO. However, it has a nonzero breaktime, thus leading to longer delay in service delivery. A typical mobile WiMAX network uses packet-switching with mostly bursty delay-tolerant data traffic. Therefore, HHO is also used in mobile WiMAX. However, for certain delay-sensitive applications, HHO must be optimized to meet the required quality of service (QoS). Recently, a large amount of research has been conducted to address the QoS of HHO in mobile WiMAX. They focused their attention on the choice of the best target BS through tracking the MS movement, analyzing the QoS information of neighbor BSs or detecting the signal level from BSs [6], [7]. In [8], data packet loss during handover was reduced via caching in the upper layer devices such as routers. To reduce delay during the handover process, Chen et al. [9] proposed a pre-coordination mechanism, which can reduce the handover latency by measuring the distance between the BS and the MS, predicting the time that handover occurs, and pre-allocating available resource for handover usage. Besides, it was shown in [9] that the handover delay could be further reduced through transmitting MAC layer messages via the IP layer. Jiao et al. [] proposed a connection identification (CID) assignment strategy to avoid conflicts of the CID assignment during the handover procedure. Hu et al. [2] applied a different handover mechanism to classified data. Both of them can decrease data delay in handover. Most previous work considered the handover performance from the viewpoint of the MAC or IP layer, but paid little attention to the physical layer. That is, physical channel quality and/or user s mobility are not taken into account. Usually, handover occurs in the boundary of BS s coverage area, where signals transmitted on the radio channel are weak and unstable. When channel quality degrades, both signalling messages and data packets are at a higher risk of being lost or corrupted. Moreover, user s mobility plays an important role in the handover process. The radio channel quality for mobile users at a higher velocity often suffers from severe /8/$2. 28 IEEE 68 ICCS 28

2 degradation due to the Dopploar frequency shift. Since the overlap area is limited between adjacent BSs, the requirement on handover latency is more stringent for users with higher mobility. In order to support full mobility (which means seamingless handover for users moving at a speed of 2 kilometer per hour or higher), these issues become severer and have to be addressed. In this work, we analyze the handover performance affected by the radio channel quality, measured in terms of the bit error rate (BER), and user s mobility in terms of the velocity by means of modeling the handover procedure. The primary handover performance in this work is considered as the probability of a successful handover, i.e., the probability that all necessary handover signaling messages are completed successfully within the required time limit imposed by user s mobility. The probability of a successful handover does not only indicate the connection-level QoS, but also play an important role in application-level QoS such as delay and/or dropping of application data packets. Based on our analysis, a simple yet effective solution is proposed to improve the successful handover probability for users with higher mobility. The rest of this paper is organized as follows. We discuss the relationship between the successful handover probability and the BER and analyze the probability at different velocities in Sec. II. Then, we propose a scheme to support high mobility using forward error correction (FEC) codes to protect the handover messages in Sec. III. Simulation results are presented in Sec. IV for performance evaluation and comparison. Finally, concluding remarks and future research directions are given in Sec. V. II. HANDOVER AND MOBILITY ANALYSIS A. Handover Procedure In the IEEE 82.6e standard, a BS periodically broadcasts the Neighbor Advertisement Message (MOB_NBR-ADV) for the identification of the network and the channel characteristics of neighbor BSs. MSs always listen to this message and collect the information about neighbor BSs. If an MS detects a received signal that becomes weaker, it will send a request message MOB_SCAN-REQ to its serving BS. Then, the BS allocates some radio resource and informs the MS to perform the scanning operation by sending message MOB_SCAN-RSP. After the scanning operation, the MS initiates the handover process by sending message MOB_MSHO-REQ to the serving BS. The BS replies to this request with message MOB_BSHO- RSP, and then the MS sends message MOB_HO-IND to inform the BS to cut all communications. In the final phase of handover, the MS continues to exchange ranging messages and re-entry messages with the target BS. The handover procedure described above is illustrated in Fig.. Thus, a successful handover is based on several handshaking messages transmitted successfully. B. Successful Handover Probability versus BER Without loss of generality, we assume that there are M(M 2) signaling messages transmitted between an MS and two BSs during the handover procedure. Let p i be the MS MOB_NBR ADV MOB_SCN REQ MOB_SCN RSQ MOB_MSHO REQ MOB_BSHO RSP MOB_HO IND Serving BS Neighoring BS Neighoring BS2 Scanning Process Connect to the target BS Figure. HO Notifaction messages Illustration of the handover procedure. probability of event A i, which indicates that the i-th signaling message is received successfully by MS or BS. By considering a scheme with no-transmission, we have P (A i )p i,p(āi) q i p i, () where i,, M. In the WiMAX handover procedure, the retransmission mechanism is used to guarantee the transmission of these signalling messages. We assume that a message is always retransmitted until it is successfully received by its peer because these signaling messages are hand shaking messages. By taking the retransmission into account, Eq.() can be rewritten as P (A i ) q j i p i ( p i ) j p i, (2) where N i, 2, 3, is the total number of the i-th message to be transmitted untill it is received successfully during a handover process. Since there are totally M signaling messages to be transmitted during the whole handover procedure, the successful handover probability is equal to P (A i ) ( p i ) j p i. (3) If p i is primarily determined by the BER of the wireless channel between a BS and an MS, we can express p i ϕ(p b (γ b ),L i ), where P b (γ b ) is the bit error probability when the received bit energy-to-noise ratio is γ b. Thus, If the i-th signaling message has a size of L i bits, we can re-write the above equation as p i [ P b (γ b )] Li, (4) where the independence of bit errors is assumed since most of radio channels are memoryless. The interleaving channel coding technique, which is widely adopted in wireless communications, also allows us to treat radio channels as memoryless ones. Consequently, the relationship between the successful 68

3 handover probability and BER of this wireless link is derived by substituting Eq. (4) into Eq. (3) as the total latency associated with the handover procedure can be written as An example of the relationship in Eq. (5) is plotted in Fig. 2. It is apparent that, the successful handover probability, P succ, decreases as the bit error probability, P b (γ b ), increases. When BER is relatively high (between 4 and 2 ), the use of retransmission improves the successful handover probability significantly. Probability of a successful handover.2. T M (N + N + + N ) T retx + M T prop, P (A i ) where T retx is the time interval of a timer between two { [ Pb (γ b )] Li} adjacent transmission of the same message, and T prop is the j [ Pb (γ b )] Li time interval for the electromagnetic wave to travel between thebsandthems. (5) We use D overlap to denote the overlap distance along the direction of the MS movement between two neighboring BSs, and the velocity of the MS. Then, the following constraint has to be satisfied if a handover operation is performed successfully: No Retransmission Retransmission Bit Error Rate Figure 2. The successful handover probability as a function of BER, where M 5, N i < 5, and L i 25 bits in this case. C. Probability of Successful Handover versus MS Velocity The above analysis indicates that the successful handover probability is a function of two important variables, i.e., the channel BER and the number of retransmissions allowed for signaling messages. In a practical WiMAX network, both of them are related to the moving velocity of an MS. The channel BER increases as the result of the Doppler effect with higher user velocity, thus demanding more retransmissions of handover signaling messages to complete the handover procedure. At the same time, a higher user moving speed demands faster switching between BSs before the MS moves out of overlapping coverage areas, which limits the maximum amount of delay time allowed for the handover procedure. Because of these two effects, the higher the MS velocity, the more rapidly the successful handover probability drops. For quantitative analysis, we assume that there is no waiting time associated with any of the handover signaling messages. This could be achieved in a light-loaded network, or by prioritizing the handover signaling packets. Under this assumption, T M < D overlap. With this constraint, the probability of a successful handover in Eq. (5) can be rewritten as, others, where and T M [ Ni ( p i) j p i ], if T M < D overlap, (6) (N i ) T retx + M T prop (7) p i [ P b (γ b )] Li. (8) The Doppler effect can be modeled to obtain the quantitative relationship between the channel BER and the MS velocity [3], [4]. The fast fading is assumed to be a Rayleigh channel under the Clarke-Jakes model. Then, the average received bit energy-to-noise ratio is given by γ b / log 2 K N 2 [N +2 Y ]+ NTs log 2 K ( ), (9) E b /N where Y N i (N i) J (2πf m T s i), N is the number of OFDM sub-carriers, T s is the duration of each K-ary QAM symbol transmitted on a subcarrier, N is the noise power, E b is the average transmitted bit energy, and f m f /c is the maximum Doppler frequency. For the QPSK modulation, the bit error probability is ( ) P b (γ b )Q 2γb. () By combining Eqs. (6) and (), we obtain the successful handover probability as a function of the MS velocity, which is plotted in Fig. 3. As expected, the probability decreases rapidly when the MS is moving at a speed higher than 5km/h. When the speed exceeds 7km/h, it is nearly impossible to have a successful handover. Thus, the handover becomes extremely challenging for a mobile WiMAX user moving at a vehicular speed even if we take the retransmission scheme. 682

4 Probability of a successful handover Velocity of mobile station (km/h) Figure 3. The successful handover probability as a function of MS velocity, where M 5, N i 5, andl i 25 bits in this case. III. SEAMLINGLESS HANDOVER SUPPORT AT HIGH MOBILITY To increase the successful handover probability of an MS at a vehicular speed, we need to establish more reliable wireless links in transmitting signaling messages besides using auto retransmission scheme. In this section, we investigate the improvement of the probability of a successful handover by adding extra protection for handover signaling messages. Forward error correction codes provide an error control technique for data transmission, where a sender adds redundant data to messages. These redundant data can assist the receiver detecting and correcting errors to futher improve wireless channel quality. During the WiMAX handover procedure, signaling messages mostly consist of small data packets whose size is less than 5 bytes. Consequently, the amount of parity-check data needed for error correction in each message is small and the resource required by error correction does not increase excessively. In comparison, the cost of the only retransmission scheme is relatively higher. In other words, this method offers an efficient solution to improve the handover performance. Different FEC schemes have different error correcting capabilities. For the same type of FEC codes, the more the redundant data, the higher the error correction capability. For a given MS velocity, we can estimate the size of FEC codes in order to achieve the desired successful handover probability. And for the sake of simple, we can use the average successful message probability, p, to replace p i in Eq. (6) as S im P (A i ) ( p i ) j p i () ( ) i M p M ( p) i M where S N i is the total number of retransmissions required to send all messages. According to the result derived in Sec. II, we have S N i D ( overlap v m MT prop ) T retx ( D overlap MT prop ) T retx. That is, we can obtain parameter S according to D overlap and and then, given P succ,thevalueof p could be easily calculated. To give an example, if the Reed-Solomon (R-S) error correcting codes are used to protect the messages, the R-S decoded message-error probability, p, in terms of the messageerror probability, p, is given by [5] p 2 k k(2 k ) 2 2 k jt+ ( 2 k ) j p j ( p) 2k j, (2) j where t N k 2 is the symbol-error correcting capacity of the code, N is the total number of bits, k is the number of bits in the original messages, and where x denotes the largest integer not exceeding x. Then, the number of redundant bits, N k, can be determined analytically. Fig. 4 shows the redundant bit size N k at different MS velocities to achieve three different successful handover probabilities: 5%, 8% and 99%. Redundant data size (bits) % 8% 5% Velocity of mobile station (km/h) Figure 4. The size of redundant data as a function of the MS velocity, where the original unprotected packet consists of k 25 bits. As shown in Fig. 4, the higher the velocity of an MS, the more the number of bits needed in an adaptive FEC scheme to achieve the target successful handover probability. For example, to achieve a probability of 8%, 2, and 28 bits have to be added into each handover signaling message when the MS is moving at a speed of 5, 7 and 9 km/h, respectively. Then, to achieve the target probability, we can adopt an adaptive FEC scheme that uses a different redundant bit size according to the MS velocity. 683

5 IV. SIMULATION RESULTS Computer simulation was conducted to evaluate the performance of the proposed adaptive FEC scheme with retransmission. We used the NS2 simulator and modified the WiMAX add-on from NIST. The parameters used in the simulation are summarized in Table I. Table I EXPERIMENTAL PARAMETERS Parameters Values Overlap distance 2 meters Bandwidth 5 MHz Carrier Frequency 2.5 GHz FFT size 52 Channel mode Rayleigh channel FEC code Reed-Solomon code Velocity 3-9 km/h The adaptive FEC scheme with retransmission calculated based on curves in Fig. 4 are used in our simulation to achieve three target successful handover probabilities, i.e., 5%, 8% and 99%. Simulations were run times for each velocity to obtain reliable statistical results. Simulation results of the successful handover probability at different MS velocities are shown in Fig. 5. It is evident that the proposed adaptive FEC scheme exceeds the three target successful handover probabilities at a wide range of velocities from 3 to 9 km/h. The local dip of these curves from from 7 to 9 km/h is primarily caused by the fact that the relationship between the size of redundant data and the vechicle speed is not linear. In this case, since a receiver can correct the maximum error of t 2 bits, we calculate the smallest even integer t as the redundant data size to meet the FEC code criterion. Probability of a successful handover target 99% target 8% target 5% Velocity of mobile station(km/h) Figure 5. Simulation results of the successful handover probability using the proposed adaptive FEC scheme. V. CONCLUSION AND FUTURE WORK The effect of the MS velocity on the handover procedure in mobile WiMAX networks was analyzed in this work. The successful handover probability drops significantly at higher MS velocity due to the degraded channel condition and limitation in MS s residential time in the overlap area of neighboring BSs. It was shown by analysis and computer simulation that the use of FEC with retransmission to protect the handover signaling messages can improve the probability of a successful handover. To achieve the target successful handover probability at different velocities, we proposed an adaptive FEC scheme that adaptively adjusts the number of redundant bits according to the MS velocity. It was confirmed by simulation results that the proposed FEC scheme with retransmission achieves the target successful handover probability efficiently. The queueing delay of handover signaling messages is however not yet considered in this work. In other words, the assumption of a light-loaded WiMAX network was adopted throughout the paper, where the network has sufficient resource to accommodate all handover requests immediately. The case of heavy-loaded networks is under our current investigation, where the effect of queueing delay should be considered explicitly. REFERENCES [] IEEE 82.6 Working Group, " IEEE Standard for Local and metropolitan area networks, Part 6: Air Interface for Fixed Broadband Wireless Access Systems, IEEE Std , May 25. [2] IEEE 82.6 Working Group, "IEEE Standard for Local and metropolitan area networks Part 6: Air Interface for Fixed and Mobile Broadband Wireless Access Systems Amendment 2, IEEE Std. 82.6e-25, Feburary 26. [3] G. P. Pollini, "Trends in handover design," Communications Magazine, IEEE, vol. 34, pp. 82-9, 996. [4] M. G. Williams, "Directions in media independent handover," IEICE Trans. Fundamentals, vol. E88-A, pp , July, 25. [5] D. J. Wright, "Maintaining QoS during handover among multiple wireless access technologies," in Proceedings of Intl. Management of Mobile Business, 27, pp. -. [6] R. Hsieh, Z. G. Zhou, and A. Seneviratne, "S-MIP: a seamless handoff architecture for mobile IP," in Proceedings of INFOCOM, 23, pp [7] L. Doo Hwan, K. Kyamakya, and J. P. Umondi, "Fast handover algorithm for IEEE 82.6e broadband wireless access system," in Proceeding of International Symposium on Wireless Pervasive Computing, 26. [8] S. Minsik, K. Hwasung, and L. Sangho, "A fast handover mechanism for IPv6 based WiBro system," in Proceedings of ICACT,26. [9] J. Chen, C.-C. Wang, and J.-D. Lee, "Pre-coordination mechanism for fast handover in WiMAX networks," in Proceedings of the 2nd International Conference on Wireless Broadband and Ultra Wideband Communications, 27, pp [] L. Chen, X. Cai, R. Sofia, and Z. Huang, "A cross-layer fast handover scheme For mobile WiMAX," in Proceedings of Vehicular Technology Conference, 27, pp [] W. Jiao, P. Jiang, and Y. Ma, "Fast handover scheme for real-time applications in mobile WiMAX," in Proceedings of ICC,27, pp [2] F. Hu and N. K. Sharma, "Priority-determined multiclass handoff scheme with guaranteed mobile QoS in wireless multimedia networks," Vehicular Technology, IEEE Transactions on, vol. 53, pp. 8-35, 24. [3] G. L. Stüber, Principles of Mobile Communication, 2 ed. Boston, MA: Kluwer Academic Publishers, 22. [4] K. K. Leung, S. Mukherjee, and G. E. Rittenhouse, "Mobility support for IEEE 82.6d wireless networks", in Proceedings of WCNC, 25 pp [5] Odenwalder, J. P., Error Control Coding Handbook, Linkabit Corporation, San Diego, CA, July 5,