Novel handover decision method in wireless communication systems with multiple antennas

Transcription

1 Novel handover decision method in wireless communication systems with multiple antennas Hunjoo Lee, Howon Lee and Dong-Ho Cho Department of Electrical Engineering and Computer Science Korea Advanced Institute of Science and Technology (KAIST) 373-, Guseong-dong, Yuseong-gu, Daejeon, Korea Telephone: , Fax: { hunju, hwlee}@comis.kaist.ac.kr, dhcho@ee.kaist.ac.kr Abstract Handover has been the most important thing for supporting the mobility and has been researched in various wireless communication systems. Handover schemes in systems with multiple antennas, the technology highlighted for 4G wireless communications to increase the capacity, have frequently been made and considered to be very important. However, handover decision method using the number of multiple antennas has not been researched. In this paper, we propose a new handover decision method using the number of detected antennas and list requirements and distinctive features for this scheme. Simulation results show that our proposed scheme and conventional scheme trigger handovers at the almost same point. Since our proposed scheme is simpler than the conventional scheme, it has the better performance in view of feedback quantity. I. INTRODUCTION Handover, the technology from one cell to another without any discontinuity, has been one of most important issues in wireless communication systems for a long time. Handover latency and packet loss degrade the quality of wireless communications. Especially, they are more important problems in the case of increasing system requirements and using new technology in these days. Consequently, in various kinds of systems, Wireless LAN (IEEE 802.), wireless MAN (IEEE 802.6e) [], Universal Mobile Telecommunications System (UMTS), UMTS-WLAN inter-working system [2] and many other systems using multi-hop or multiple-inputmultiple-output (MIMO), many handover schemes have been studied. Above all, many handover methods of systems with MIMO, technology highlighted for 4G wireless communication to increase capacity and reliability of wireless by transmit diversity (TD) gain or spatial multiplexing (SM) gain [3], have been researched and proposed. In this paper, we will introduce one conventional scheme with MIMO, Soft handoff for OFDM [4]. In this reference, they used the handover decision method not considered the number of antennas. Moreover, the handover decision method using the number of antennas in system with multiple antennas has never been studied so far. If the number of detected antennas from Base Stations (BS) is considered at handover decision, we will anticipate a new handover scheme which has the better performance by simplifying the channel reporting messages. This paper is structured as follows: In section II, we will survey the decision method of current SINR-based handover and the handover scheme of systems with MIMO. In section III, novel handover scheme, decided by detected number of antennas, is proposed and several methods for implementation are explained. In addition, the reduction of feedback quantity is shown. In section IV, we will analyze the change of the number of detected antennas using mathematical model. Section V discusses experimental results by simulations. Conclusion and further works are given in section VI. PILOT STRENGTH T_ADD T_DROP II. CONVENTIONAL SCHEMES PILOT IN PILOT IN NEIGHBOR SET CANDIDAT E SET Fig.. PILOT PA TIME t0 t t2 t3 t4 PILOT IN ACTIVE SET PILOT IN NEIGHBOR SET Management of active sets in conventional handover scheme Handover technology is one of most important things for supporting the movement in wireless systems. Generally, it has been operated by comparison of Signal to Interference U.S. Government work not protected by U.S. Copyright

2 TIME TRAFFIC DATA SCATTERED PLOT NULL FREQUENCY BS BS2 BS3 Fig. 2. Soft handoff for OFDM based on frequency division Fig. 3. Operation overview of proposed scheme and Noise Ratio(SINR). Figure describes the conventional handover scheme for the management of candidate sets, which can immediately be connected by handover, in case of systems with single antenna [5]. If the pilot strength from one neighbor BS is more powerful than threshold(t ADD), the BS will be changed into candidate set from neighbor set. If the pilot strength is maintained for t t 0, the BS could be included in the active set, communicating with the MS. If the pilot strength from one of active sets is less than threshold(t DROP), the MS will start Handover drop timer. If Handover drop timer expires, the BS will be included in the neighbor set. In other words, Serving BS or MS has to compare with the pilot strengths of BSs belonged to active sets, and make a decision to handover to neighbor BS having most strong pilot strength. We define that this method is called the conventional scheme in this paper. Figure 2 describes the more practical example, the soft handover scheme using division in frequency domain for OFDM systems using multiple antennas [4]. In the right figure, when soft handover happens, packets are duplicated and transmitted by multiple antennas of other active soft handover- BS. Other figure shows the diagram of frequency sub-band structure according to embodiment of this scheme. In this scheme, only average Carrier to Interference Ratio(CIR) of antennas is considered during a decision of handover, and the detected number of antennas is not reflected. III. PROPOSED NOVEL HANDOVER SCHEME A. Algorithm of Proposed Scheme Although a number of studies have been made on the handover schemes, the number of detected antenna has not been considered. If we use the number of antennas during the handover decision, we will expect a new algorithm, which is simpler than conventional schemes. Therefore, we propose a novel handover scheme, which could not be used in the systems with a single antenna. The whole overview of proposed handover scheme is described in Figure 3. There are two BSs, which have four transmit antennas individually. In the first picture, MS communicates with BS by four antennas. As MS moves from BS to BS2, the number of detected antennas from BS decreases and the number of detected antennas from BS2 increases. Second picture shows that MS detects the same number of antennas of two BSs. Then, handover process will start if MS keeps going to BS2. Third picture describes a point of handover trigger, when MS is sensing more antennas of BS2 than BS. The diagram of new handover process by MS requests is described in Figure 4. Firstly, MS measures the pilot signal strength of each antenna at each BS and computes the CIR of each antenna(j) for each BS(i). No. of detected antennas(i) which has the information about the number of available antennas at each BS(i) is refreshed by checking whether the CIR of each antenna is more than the threshold of CIR. If CIR ij is stronger than CIR th, it will be added to No. of detected antennas(i). After selecting Best BS which has the most largest number of detected antennas having pilot strength exceeding threshold, MS transmits handover initiation message to the serving BS. Most important feature in proposed scheme is a fresh handover decision method without conventional method based on the comparison of the SINRs between BSs. It is a new criterion for the handover decision, using the number of detected multiple antennas. Other feature is to reduce MAC overhead reported at handover by reporting only the number of antennas instead of the CINR. This is explained in detail at section D. B. Method to Determine Threshold of CIR There are two methods how to determine the threshold of CIR. The first is theoretical method computing the CIR by normalized distance. The following equation shows the relation of CIR vs. the normal distance(r), which is the distance between BS and MS divided by the coverage of BS,

3 CIR = r α 2p[( 3 r ) α +( 3+r2 3 )α +( 3+r 2 + 3r )α ]. (),where pathloss exponent α is 4, traffic load is p [6]. Only pathloss in channel environments is considered and fading is not considered. We determine the threshold value using this equation in the simulation. For example, when the normal distance(r) is and traffic load is 0.5, the threshold of CIR is.5db. Using Minimum-Decodable-CIR, in which each user has independent threshold, is the second method. For the best suited handover, every user in different circumstances can calculate its the threshold of CIR by decoding some codes. Needless to say any more, this supports the best fitted handover, which can look for accurate handover points. It will be simulated by further works. C. Method to Distinguish pilots from Antennas of BSs The method which identifies pilots from each antenna of every BS has to be required. For distinction, we have to consider the multiplexing with the orthogonal pilots. There are several multiplexing examples, time division multiplexing (TDM), frequency division multiplexing (FDM), code division algorithm (CDM), and etc. Figure 5 shows the frame structures which have the distinctive pilots in the case of using TDM(scheme ) and TDM with FDM(scheme 2). In scheme, we can identify the antenna pilot without interference, because BS sends only one pilot at each time slot. Moreover, in scheme 2, pilots of each antenna are transmitted by individually different frequency. D. Reduction of Feedback Quantity As we use the proposed scheme, we can reduce the quantity of feedback information for deciding target BS of handover. We can see that MAC overhead reported for handover are reduced by reporting only the number of antennas instead the CIR values of the antennas. As a consequence, proposed scheme can enhance the effective throughput. For example, in case of conventional handover scheme, the MS sends MOB MSHO-REQ and MOB ASC-REP which have each Fig. 5. Frame structures for identifying antennas s pilot eight bits for the average CIR of each antenna in the BSs included in the MS s handover active set in the 802.6e systems []. If the number of target BSs increases, reporting quantity is steeply larger. However, when we use our proposed scheme, we can use only each two bits for the number of detected antennas of each BS. The total quantity of feedback reduction is given as follows L s = N m N tbs N ant (L con L pro ) (2) Here, L s, N m, N tbs, N ant, L con, and L pro represent the number of total bits reduced in the proposed scheme than conventional scheme, the number of feedback messages, the number of target BS, the number of transmitted antennas in a BS, the number of bits for reporting of the message in conventional scheme and the number of bits for reporting of the message in proposed scheme, respectively. According to this equation, as more target BSs and more frequently reported message are used, there is more reduction of bits and it is more useful. As more neighbor BSs, there is more differences. The proposed scheme is more useful in case of systems using the messages which are more frequently reported. Fig. 4. Algorithm of proposed scheme IV. ANALYSIS ON NUMBER OF ANTENNAS In this section, we analyze the variance of the number of detected antennas. We assume that the pilot strength of each antenna is affected by pathloss and long-term fading. In many experiments, it is shown that long-term fading obeys log-normal distribution. The probability probability density function (pdf) of received power variability level according to long-term fading is given by ) (lnγ γ)2 f(γ) = exp ( 2πσγ 2σ 2,γ 0 (3)

4 Number of antennas Distance from old BS And there are four transmitting antennas and two receiving antennas. Then, pathloss model is ITU-R s model and fading is Jake s model. Table I describes the simulation models for implementation. TABLE I SIMULATION MODEL Target system MIMO-OFDMA Frequency 2.3GHz No. of subcarriers 024 Channel type pedestrian A No. of Tx antennas 4 No. of Rx antennas 2 Pathloss model 37.6 log0(r)+6.62(r in meter) Fading Channel model Jake s channel fading model Threshold of CIR normalized distance(0.975) Fig. 6. Number of detected antennas vs. distance between user and BS In this equation, x = logγ is normal distribution with mean γ = E[x] and variance σ 2 = Var(x) [7]. For the convenience, we assume the summation of interferences is not affected by log-normal fading. Accordingly, CIRs of antennas of each BS obey log-normal distribution with the same mean, which is calculated by (). Then, as MS goes away serving BS, the detection probability of each antenna is calculated by the cumulative density function (CDF) considering CIR th as follows P d (r) = g(p)dp (4) CIR th In this equation, P d, r, CIR th and g(c) represent the detection probability of each antenna, the distance from BS, the threshold of CIR and the pdf of CIR according to lognormal distribution, respectively. Using this probability, we can estimate the detected number of antennas by multiplying the number of transmitting antennas as follows B. Simulation Results and Discussion Figure 7 illustrates the number of detected antennas as the distance from BS increases in case of proposed scheme. We plot the number of detected antennas from BS and BS2. As distance between the user and BS is increasing, the number of detected antennas of BS is decreasing and the number of detected antennas of BS2 is increasing linearly. They cross at about 950m. And, we can guess that it is about a point of handover trigger. For the direct comparison, we simulate the conventional scheme, too. Figure 8 shows the SINR curve when the distance from BS to the user increases in case of conventional scheme. As the distance from BS is increasing, the average SINR from BS is exponentially decreasing and the average SINR from BS2 is exponentially increasing. Like proposed scheme, the handover occurs at about 950m. To compare the results of proposed scheme with those of conventional scheme in detail, we plot the simulation results of two schemes using normalized value in the consideration of roughly equal scales in N d ant (r) =N ant P d (r) (5) In this equation, N d ant and N ant represent the number of detected antennas and the number of antennas in a BS, respectively. When N ant is 4 and the deviation σ between antennas is 4, the variation of the number of detected antennas can be shown in Figure 6. This Figure shows the characteristic that the number of detected antennas decreases at the handover region. V. PERFORMANCE EVALUATION A. Simulation Environments In this section, we compare proposed scheme with conventional scheme by computer simulation. The system used in this simulation is MIMO-OFDMA system. We assume only one neighbor BS, and use the threshold of CIR, and fixed normalized distance, 0.975, calculated by theoretical results. Fig. 7. Number of detected antennas vs. distance between user and BS for proposed scheme

5 from proposed scheme is different from BS selected by conventional scheme using average SINR, we will investigate which one has better performance, too. VII. ACKNOWLEDGEMENT This research was supported in part by University IT Research Center Project. REFERENCES Fig. 8. SIR vs. distance between user and BS for conventional scheme Figure 9. This figure describes that proposed scheme triggers handover at a little early point than conventional scheme. We can see from this figure that the difference of location where handover occurs for two schemes is very small compared with the conventional handover region. [] IEEE 802.6e, IEEE Standard for Local and Metropolitan Area Networks - Part 6: Air Interface for Fixed and Mobile Broadband Wireless Access Systems - Amendment for Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands, Feb [2] Hyun-Ho Choi, et al., A Seamless Handoff Scheme for UMTS-WLAN Interworking GLOBECOM 04. IEEE, vol.3, pp , Dec [3] L. Zheng and D.Tse, Diversity and multiplexing: a fundamental tradeoff in multiple-antenna channels, Information Theory, IEEE Transactions on Vol. 49, No. 5, pp , May [4] MA, Jianglei, et al., Soft handoff for OFDM, International Patent(WO 03/08938 A), Oct [5] Blakeney, II et al., Mobile station assisted soft handoff in a CDMA celluar communications sysytem, United State Patent(5,640,44),Jun [6] Chansu Hwang, et al., Analysis and comparison in MC-CDMA and Frequency Hopping OFDMA, Telecommunications Review, vol. 2, no 4 pp , Jan [7] Charalambous C.D. and Menemenlis N., General Non-Stationary Models for Short-Term and Long-Term Fading Channels, EUROCOMM 2000, pp , May Fig. 9. Normalized values vs. distance VI. CONCLUSIONS In this paper, we proposed a novel handover scheme using the number of detected antennas in wireless communication system with MIMO. It brings the diversification of handover decision unlike the conventional scheme and the reduction of MAC overhead. As results of simulation, proposed scheme and conventional scheme triggered handover at the almost same point. Also, our proposed scheme is simpler than conventional scheme and can diminish MAC feedback quantity. As further works, the method, which looks for the decodable CIR value for each user, will be researched. When BS selected