Hybrid Station Aided Coexistence Scheme between Wireless PANs and Wireless LAN

Transcription

1 578 PAPER Special Section on Wideband Systems Hybrid Station Aided Coexistence Scheme between Wireless PANs and Wireless LAN Fumihiro INOUE a), Student Member, Takayuki NISHIO, Member, Masahiro MORIKURA, Fellow, Koji YAMAMOTO, Senior Member, Fusao NUNO, Member, and Takatoshi SUGIYAMA, Senior Member SUMMARY The problem of coexistence between IEEE g based wireless LANs (WLANs) and IEEE based wireless personal area networks (WPANs) in the 2.4 GHz band is an important issue for the operation of a home energy management system (HEMS) for smart grids. This paper proposes a coexistence scheme that is called a Hybrid station aided coexistence (HYSAC) scheme to solve this problem. This scheme employs a hybrid-station (H-STA) that possesses two types of network device functions. The scheme improves the data transmission quality of the WPAN devices which transmit energy management information such as power consumption. The proposed HYSAC scheme employs WLAN control frames, which are used to assign WPAN system traffic resources. Moreover, we propose a coexistence method to achieve excellent WLAN throughput where multiple WPANs coexist with a WLAN. We theoretically derive the performance of the proposed scheme by considering the QoS support in WLAN and show that the results of the simulation and theoretical analysis are in good agreement. The numerical results show that the HYSAC scheme decreases the beacon loss rate of WPAN to less than 1% when the WLAN system consists of 10 STAs under saturated traffic conditions. Furthermore, the WLAN throughput of the proposed synchronization method is shown to be 30.6% higher than that of the HYSAC scheme without synchronization when the WLAN that consists of 10 STAs coexists with four WPANs. key words: coexistence, IEEE , IEEE , ZigBee, WLAN, WPAN,WSN,QoS 1. Introduction Machine-to-machine (M2M) networks, which comprise of power meters and various sensors, are currently attracting considerable attention for the realization of smart grids and smart communities. The wireless access network for a home energy management system (HEMS) is generally realized by networks based on the IEEE [1] and [2] standards; they are called wireless LAN (WLAN) and a wireless personal area network (WPAN), respectively. In the IEEE standard, the wireless device has two modes, a beacon-enabled mode and a non-beacon-enabled mode. Kim et al. have analyzed the performance of the nonbeacon-enabled mode [3]. In the beacon-enabled mode, devices can reduce energy consumption by going into sleep mode during its inactive periods. Manuscript received April 8, Manuscript revised September 4, The authors are with the Graduate School of Informatics, Kyoto University, Kyoto-shi, Japan. The authors are with NTT Access Network Service Systems Laboratories, NTT Corporation, Yokosuka-shi, Japan. This paper was presented in part at IEEE SmartGridComm a) info14@imc.cce.i.kyoto-u.ac.jp DOI: /transfun.E98.A.578 Fig. 1 band. Wireless channels of IEEE g and IEEE at 2.4 GHz WLAN and WPAN devices operate at the same frequencybandsuchas2.4ghzasshowninfig.1andat the same location. This scenario generates a serious interference problem in the WPAN transmissions. Myoung et al. have analyzed WLAN performance in the presence of a WPAN [4]. WLAN is resistant to the interference from WPAN. On the other hand, Shin et al. have analyzed the packet error rate of a WPAN with interference from a WLAN [5]. In [5], the packet error rate of WPAN is more than 20% when there are five WLAN STAs and the distance between WLAN STAs and WPAN device is 6 m. It is difficult for WPANs to be employed for applications that are required reliability. Frequency channel allocation management systems are conventionally employed to solve this problem [6] [8]. However, housing complexes and offices generally use three channels of the WLAN at the 2.4 GHz band. This leads to a shortage of the number of channels for the WPAN and causes frequency channel overlapping between WLANs and WPANs, which generates severe frequency interference problems. To solve these frequency channel overlapping problems, Shin et al. proposed an active channel reservation for coexistence (ACROS) scheme [9] that exploits beacon-enable mode. However, the ACROS scheme has a problem when the WLAN has bidirectional traffic. It is the degradation of the probability of successful WPAN beacon frame transmission. As a result, WPAN devices can not transmit their frames during a superframe. Moreover, the ACROS scheme does not consider multiple WPANs in a WLAN cell. If we apply multiple WPANs to the ACROS scheme, the WLAN throughput may be severely degraded. We propose a hybrid station aided coexistence (HYSAC) scheme for multiple WPANs to solve the above mentioned problem. The proposed scheme employs a hybrid station (H-STA) that possesses two functions. The two Copyright c 2015 The Institute of Electronics, Information and Communication Engineers

2 INOUE et al.: HYBRID STATION AIDED COEXISTENCE SCHEME BETWEEN WIRELESS PANS AND WIRELESS LAN 579 functions are a personal area network-coordinator (PAN-C) of the WPAN and a station (STA) of the WLAN. In order to realize high-quality communication for the WPANs and WLAN, we optimize the medium access control (MAC) parameters of the H-STA. We derive the WPAN protection failure probability and WLAN throughput performance of the proposed HYSAC scheme by both theoretical analysis and computer simulation. The performance evaluation shows that the proposed HYSAC scheme outperforms the conventional ACROS scheme. Moreover, we propose HYSAC with a synchronization method to solve the problem that WLAN throughput is degraded due to multiple WPANs. Numerical results show that the WLAN throughput of the proposed method outperforms that of HYSAC without synchronization. The rest of this paper is organized as follows. The conventional coexistence scheme is presented in Sect. 2. The proposed coexistence scheme is described in Sect. 3. Section 4 presents the theoretical analysis. Section 5 shows some numerical results. Finally, we present some concluding remarks in Sect. 6. Fig. 2 Timing diagram of the conventional ACROS scheme to obtain a WPAN superframe period, which is contention-free from WLANs. 2. Conventional Scheme: ACROS Shin et al. proposed a protection scheme for a beaconenabled WPAN against a WLAN called ACROS. The key idea of the ACROS mechanism is to reserve a wireless channel for the superframe of the WPAN. ACROS exploits the WLAN virtual carrier sensing with the request-to-send (RTS) frames from the access point (AP) in the WLAN. After receiving the RTS frame, the WLAN STAs defer their transmissions by setting a network allocation vector (NAV). Shin et al. proposed incorporating both the WLAN AP function and the WPAN PAN-C function into ACROS device to implement the ACROS. Figure 2 shows the timing diagram indicating the time for the ACROS system to obtain a WPAN superframe period that is contention-free from the WLANs. The ACROS device begins carrier sensing for an RTS frame at a certain period before the transmission of the WPAN beacon frame. For descriptive purposes, we define this period as the WPAN protection attempt period. In ACROS, an RTS frame is transmitted just once before the beacon frame transmission of the WPAN. If the RTS frame does not collide with the other frames and all STAs can receive it correctly, they set the NAV and defer their data frame transmissions. Therefore, these WPAN devices are able to communicate with each other without interference from the WLAN. In a practical system, the traffic would be bidirectional. In bidirectional traffic, the RTS frame could collide with the data frame transmitted by the STAs. If the collision occurs, the ACROS device is unable to detect the collision state, and a WPAN beacon frame may collide with the WLAN data frames. Moreover, the performance of ACROS is evaluated only by experiments. Thus the operating principle of ACROS is not clarified. QoS control of WLAN is not considered when the WLAN has bidirectional traffic. Besides, Fig. 3 Timing diagram of multiple WPAN beacon and superframe transmissions by using ACROS scheme. the incorporation of the WLAN AP and WPAN PAN-C imposes a constraint on the locations of the WPANs. In addition, the conventional ACROS scheme does not consider multiple WPANs in a WLAN cell. We explain the behavior of an ACROS device in the coexistence between a WLAN and multiple WPANs. We assume that each WPAN uses a different frequency channel and that these WPAN channels completely overlap with the specified WLAN channel. The ACROS device sets multiple NAV periods for the WLAN to acquire multiple superframe periods without interference. Figure 3 shows the timing diagram of the ACROS scheme in this case. When the superframe periods do not overlap as shown in Fig. 3(a), WLAN throughput decreases significantly according to increase of NAV periods. When parts of the superframes overlap with one another, as shown in Fig. 3(b), WPAN 2, which plans to transmit a beacon frame later, cannot acquire the superframe period without interference since the ACROS device cannot transmit an RTS frame for a superframe of WPAN 2 because of the superframe transmission of WPAN Proposed Scheme: Hybrid Station Aided Coexistence (HYSAC) The proposed HYSAC scheme represents a coexistence scheme between a WLAN and multiple WPANs that operate in the same frequency band. The HYSAC scheme consists of two processes: one is acquisition of a superframe period

3 580 Fig. 4 Structure of H-STA. Fig. 6 Timing diagram of the proposed HYSAC scheme to obtain a WPAN superframe period, which is contention-free from WLANs. Fig. 5 Coexistence model for IEEE g and IEEE in the 2.4 GHz band. Each H-STA is located within the carrier sense area of AP. The number of NDs that communicates with one H-STA is assumed to be approximately ten. without interference from WLAN nodes, and the other is the synchronization among multiple WPANs in the WLAN. The details are presented in Sects. 3.1 and 3.2. The proposed HYSAC scheme employs the H-STA. The H-STA is implemented by the integration of an IEEE based PAN-C and an IEEE g based STA as illustrated in Fig. 4, while the conventional ACROS is implemented by integrating a WLAN AP and a PAN-C. In Fig. 4, the module at the left is a WLAN STA part, and the module at the right is a PAN-C part. In addition, H-STA possesses some functions, which are a synchronization function between the STA and PAN-C parts with a timing synchronization function (TSF) timer information, WPAN protection controller, RTS/clear-to-send (CTS) frame controller, a table of WPAN information, and a determination part of the representative H-STA that acquires superframe periods. In the proposed scheme, a mechanism for the RTS/CTS frames is exploited to detect the collision state of an RTS frame and suppress the data frame transmissions from the destination of the RTS frame. This achieves a WPAN communication quality higher than that achieved by ACROS. In addition, multiple WPANs can be located independently of the WLAN AP. Figure 5 shows the coexistence model of a WLAN and WPANs. The WLAN includes one AP, n STAs, and r H-STAs. Each H-STA is located within the carrier sense area of the AP. All the WLAN STAs and the AP transmit data frames by using the basic access mechanism of the IEEE DCF. We assume that the WLAN has bidirectional traffic and that the downlink traffic is larger than uplink traffic [10]. To guarantee that downlink throughput is greater than uplink throughput, the CWmin value of the AP is less than that of the STAs. The WPAN consists of the H- STA (which includes the PAN-C function) and some of the WPAN network devices (NDs) and operates in the beaconenabled mode. The number of NDs that communicates with one H-STA is assumed to be approximately ten. 3.1 Acquisition of Superframe Period without Interference from WLAN Figure 6 shows the timing diagram of the proposed HYSAC when r = 1, indicating the method to obtain a WPAN superframe period without the interference of the WLAN. The PAN-C periodically transmits a WPAN beacon frame to NDs with the same PAN ID. WPAN protection controller sends an RTS frame request to the WLAN part of the H- STA via the RTS/CTS frame controller before the WPAN beacon transmission. The H-STA attempts to transmit an RTS frame with the destination address of the AP through the carrier sense multiple access with collision avoidance (CSMA/CA) protocol in a certain period T PAP (a WPAN protection attempt period) before WPAN beacon frame transmission. If the RTS frame collides with other data frames and the AP cannot receive it, the H-STA detects the collision by waiting for a CTS timeout period. After collision detection, the H-STA transmits the RTS frame again by using the random binary exponential backoff algorithm of the CSMA/CA. After the AP receives the RTS frame correctly, it transmits a CTS frame back to the H-STA. If the H-STA receives the CTS frame correctly, all the WLAN nodes defer their transmission by setting a NAV period, and the AP turns into a receive state. This procedure should be completed within the WPAN protection attempt period. When

4 INOUE et al.: HYBRID STATION AIDED COEXISTENCE SCHEME BETWEEN WIRELESS PANS AND WIRELESS LAN 581 Fig. 7 Timing diagram of WPAN beacon and superframe transmissions when the proposed synchronization method is applied to multiple WPANs. a blocking event between an H-STA and the AP occurs, the H-STA cannot receive the CTS frame from the AP. In this case, WPAN protection controller informs the PAN-C of the H-STA that the procedure failed, and PAN-C does not perform WPAN communication. 3.2 Synchronization Method between Multiple WPANs We propose a synchronizing method of beacon transmissions between multiple WPANs. This synchronization method improves WLAN throughput thanks to decrease of a total NAV period in the WLAN. Figure 7 shows a timing diagram of WPAN beacons and superframe transmissions when the proposed method is applied to multiple WPANs. Multiple WPANs can transmit their own beacons and get superframes in the same period when beacon transmission instants of multiple WPANs are synchronized. In this case, each WPAN uses the different radio frequency channel. This synchronization method can be applied to up to four WPANs in one WLAN cell because there are four orthogonal WPAN channels in 22 MHz bandwidth as shown in Fig. 1. The required WPAN beacon timing error from the beacon timing of the representative H-STA is within 34 μs, which corresponds to a DCF interframe space (DIFS) period, to synchronize each WPAN. This is because the WLAN nodes defer their transmission for 34 μs from the end of the NAV period. When the beacon interval is equal to 100 ms, the required frequency stability of the H-STA s clock is within ± 340 ppm. The commercially available clocks have much better stability performance. The WLAN TSF timer of an H-STA is synchronized with that of the other H-STAs by receiving WLAN beacon frames from the AP. Simultaneously, each H-STA synchronizes the WPAN timer with the WLAN TSF timer. We assume that the residual timing error of the synchronization between the WLAN and WPAN parts is small enough to satisfy the above requirement of the beacon timing error among all WPANs. In the proposed method, only one H-STA among multiple H-STAs transmits the RTS frame to avoid collisions between RTS frames. We call it the representative H-STA Requirements for the Representative H-STA When multiple H-STAs are synchronizing their beacon transmission instants, the H-STA whose beacon interval is Fig. 8 Timing diagram of beacon interval based on IEEE The H-STA whose beacon interval is the shortest of all, operates as the representative H-STA. the shortest of all H-STAs, operates as the representative H- STA. Figure 8 shows a timing diagram of beacon transmissions based on the IEEE standard. The beacon interval is given by ms 2 BO (0 BO 15). Therefore, the beacon transmissions with the shortest WPAN beacon interval overlap with the beacon transmissions of the other WPANs once their instants of beacon transmissions are synchronized. As a result, the WLAN nodes defer their transmissions during the periods of all beacons and superframes of WPANs in the WLAN cell. The representative H-STA sets a sufficiently long NAV period for the longest superframe of WPANs Procedure of Synchronizing WPANs In order to realize synchronization of beacon instants of WPANs, H-STAs determine which H-STA operates as the representative H-STA in a distributed manner and the representative H-STA informs other H-STAs of the next beacon transmission instant. We explain procedures how to decide the representative H-STA among H-STAs, and synchronization of beacon instants. We assume that each H-STA has a table that lists the beacon orders (BO), superframe orders (SO), operating frequency channels, and MAC addresses of all H-STAs in the same WLAN cell to determine the representative H-STA. The BOs and SOs values determine the WPAN beacon interval and the superframe length, respectively. Figure 9 shows the flowcharts that define operations of H-STAs to synchronize their beacon instants. When a new H-STA joins a WLAN cell, the H-STA executes the process (A) as shown in Fig. 9 and stays at the steady state. In the process (A), the H-STA broadcasts the table of own WPAN information, by the function shown in Fig. 4, that contains its beacon interval, the superframe length, the operating frequency channel of WPAN, and the MAC address by using a WLAN data frame of the WLAN channel via the AP, and then the representative H-STA receives it and adds the information to its table and broadcasts the information of the table to all H-STAs via the AP. As a result, the H- STA that takes part in the WLAN obtains information of all H-STAs. The function of the representative H-STA determination shown in Fig. 4 is to receive the table of all WPAN in-

5 582 Fig. 9 Flowcharts that define operations of H-STAs to synchronize their beacon instants. formation in each H-STA. It chooses a vacant WPAN channel and determines in a distributed manner whether the H- STA is a representative or not according to the table. The representative H-STA informs the other H-STAs of the next beacon transmission instant. The other H-STAs adjust their beacon transmission instants to the instant informed by the representative H-STA. In a steady state, the tagged H-STA executes the process (B) as shown in Fig. 9 when other H-STAs join in the WLAN, and executes the process (C) when it is considered that the representative H-SAT has left the WLAN. By executing these processes, the tables of H-STAs are renewed and the representative H-STA is determined in a distributed manner. The new representative H-STA informs the others of the instant of the next beacon transmission. After the above described processes, the representative H-STA sends the longest superframe period to WPAN protection controller, as shown in Fig. 4. WPAN protection controller requests the WLAN part to transmit an RTS frame before the WPAN beacon transmission to defer transmissions by WLAN nodes during the WPAN transmission. The other H-STAs transmit their beacons and obtain their superframe periods at the same instants without transmitting an RTS frame. 4. Theoretical Analysis of HYSAC In this section, we derive the superframe acquisition performance by the representative H-STA and the WLAN throughput performance using the proposed synchronization method. The former is derived from the coexistence model with a single WPAN, and the latter is derived from the coexistence model with multiple WPANs. 4.1 Coexistence Model with Representative H-STA In this section, we assume that a WLAN and a single WPAN coexist to evaluate the performance of the superframe acquisition by the HYSAC scheme. In the proposed system, the WLAN part of H-STA does not transmit any frames except RTS frames to protect the WPAN superframe. Therefore, Fig. 10 Relationship between T BI (beacon interval), T NAV (allocated NAV period), T ST (required time from the beginning of RTS frame generation to successful CTS frame reception), T PAP (WPAN protection attempt period), and T SD (superframe duration) in the HYSAC scheme. the WLAN nodes, including the AP and STAs except the H-STA, remain at a steady state of the stochastic process. On the other hand, the H-STA is not in a steady state of the stochastic process when RTS frames are generated in the H- STA. At this moment, by applying the analytical model for the IEEE DCF with priority control [11] to WLAN nodes, we can derive the values of steady state distribution Π k (s, j) of respective WLAN nodes. Π k (s, j) represents the steady state distribution of WLAN node k where the backoff stage value equals s andthebackoff counter value equals j at the start of a cycle. One cycle is defined as a period which starts after a DIFS and ends at the next DIFS [11]. It is expected that RTS frames of H-STA tend to collide with data frames of WLAN nodes that have small CWmin value. We derive the performance of the proposed scheme by considering WLAN nodes with QoS support in a steady state and the H-STA in a non-steady state. Let T ST denote the elapsed time from the beginning of RTS frame generation in the H-STA to successful CTS frame reception during the WPAN protection attempt period T PAP, as shown in Fig. 10. Then, the WPAN protection failure probability P PF can be expressed as the probability that T ST is greater than T PAP. In order to derive the probability distribution of T ST, we express T ST as a function of l and λ. l denotes the number of collisions between RTS frames of H-STA and data frames of WLAN nodes during this period. λ denotes the number of data frame transmissions by WLAN nodes dur-

6 INOUE et al.: HYBRID STATION AIDED COEXISTENCE SCHEME BETWEEN WIRELESS PANS AND WIRELESS LAN 583 ing the backoff procedure of the H-STA. The period of data frame transmission is dominant during T ST. We can neglect T Backoff which is the expected backoff period because the period of data frame transmission is much longer than the slot time σ. Byusingl and λ we can derive T ST (l,λ) as follows, T ST (l,λ) = T DATA (l + λ) + T Backoff +T RTS + T SIFS + T CTS T DATA (l + λ) +T RTS + T SIFS + T CTS, (1) where T DATA, T RTS, T SIFS,andT CTS denote the data frame transmission duration, the RTS frame duration, the SIFS period, and the CTS frame duration, respectively. T DATA denotes the expected period of DATA frame transmission. It is expressed as T DATA = P succ T succ + (1 P succ )T coll, (2) where T succ represents the successful data frame transmission period which is composed of a data frame duration, an SIFS period, an ACK frame duration, and a DIFS period, T coll denotes a collision period which is composed of a data frame transmission period and a DIFS period, P succ denotes the successful data frame transmission probability, which is derived from the steady state distribution Π k (s, j). We derive P PF (T PAP ) that is WPAN protection failure probability as a function of the WPAN protection attempt period. Let R(l,λ) denote the probability that RTS frame collision occurs l times and data frame transmission of the WLAN nodes occurs λ times before successful RTS frame reception by the AP. Then, P PF (T PAP ) can be derived from the summation of R(l,λ) where the combination (l,λ) satisfies the condition that T ST (l,λ) is greater than T PAP. P PF (T PAP ) is expressed as P PF (T PAP ) = R(l,λ), (3) l,λ A where A = {l,λ T ST (l,λ) > T PAP }. (4) In what follows, we define time slot as either the constant time interval σ, or the variable time interval between two consecutive backoff counter decrements [11]. To derive R(l,λ), we assume that the frame transmission probability at each time slot during the backoff procedure has a constant value τ according to Bianchi s theoretical analysis [12]. τ is derived as follows, E[x] = CWmax H-STA i=0 i(1 τ) i τ 1 τ τ, (5) 1 τ 1 + E[x], (6) where E[x] denotes the expected value of the number of idle time slots elapsing after a DIFS period. CWmax H-STA is the maximum value of a contention window that an H-STA can have. E[x] can be derived from the steady state distribution Π k (s, j). By using this approximations, we can use a binomial distribution with parameters m and τ during T ST. m denotes the total number of time slots corresponding to T ST. Accordingly, the probability f m (λ) that the WLAN nodes transmit data frame λ times in total m time slots is obtained from the following binomial distribution, ( m f m (λ) = τ λ) λ (1 τ) m λ,λ {0, 1,, m}. (7) By using Eq. (7), R(l,λ) is derived as follows, m max (l) R(l,λ) = P(l) P slot,l (m) f m (λ), (8) m=0 where P(l) denotes the probability that the RTS frame collides l times with the data frames transmitted by the WLAN nodes, and issuccessfullyreceivedat (l + 1)th transmission, m max (l) denotes the maximum value of m in case that the RTS frames collide l times. Let CW i denote the contention window size of the H-STA at ith transmission. By using CW i, m max (l) is calculated as l+1 m max (l) = CW i, (9) i=1 where CW i = (CWmin H-STA + 1) 2 i 1. (10) P slot,l (m) represents the probability that total m time slots are elapsed before successful RTS frame reception by the AP under the condition that the RTS frame collision occurs l times. P slot,l (m) is derived as follows, 1 0 m CWmin, CWmin + 1 l = 0, 0 otherwise P slot,l (m) = m 1 P slot,l 1 ( j) l 1. CW l j=m CWl+1 (11) To derive P(l) which denotes the probability that the RTS frame collides with the data frames transmitted by the WLAN nodes l times and is successfully received at (l+1)th transmission, let P col (i) denote the collision probability of the ith RTS frame transmission of the H-STA. As mentioned at the former paragraph, we assume that the data frame transmission probability at each time slot by WLAN nodes has a constant value τ. Then, P col (i) is derived as P col (i) = τ, i. (12) By using P col (i), P(l) can be expressed as P(l) = P col l (1 P col ). (13) As described above, protection failure probability P PF can be calculated from Eq. (3). Further we derive the system throughput performance of the WLAN. The system throughput S total that consists of S down and S up of the WLAN, which does not contain the H-STA, is derived as S total = S down + S up (14) = P succ(ap) PSIZE AP E[cycle] (15)

7 584 + n P succ(sta) PSIZE STA, (16) E[cycle] where PS IZE AP and PS IZE STA are the MAC service data unit sizes of the AP and STAs, respectively. E[cycle] is the expected value of one cycle duration, and P succ (k) denotes the probability that WLAN node k successfully transmits in acycle. P succ (k) is defined in [11]. P succ (k) depends on the number of STAs n and monotonously decreases according to the increase in n. E[cycle] is expressed as E[cycle] = E[x]σ + T DATA, (17) where σ denotes the slot time. The throughput degradation ratio D is calculated by using the NAV period T NAV that the H-STA successfully allocates. The throughput degradation ratio D is the ratio of T NAV to T BI which denotes the beacon interval as shown in Fig. 10. D is expressed as D = T NAV T BI = T PAP E[T ST ] + T SD T BI, (18) where T SD is the superframe duration, and E[T ST ]istheexpected value of T ST assumed that the H-STA successfully allocates NAV period. E[T ST ] is calculated as follows, E[T ST ] = T ST (l,λ) l,λ A R(l,λ) 1 P PF (T PAP ), (19) where A is a complementary set of A. Taking into consideration of D and P PF (T PAP ) which denotes the probability that no NAV periods are assigned, the WLAN throughput performance of HYSAC S HYSAC is derived as S HYSAC = S total [1 (1 P PF (T PAP )) D]. (20) 4.2 WLAN Throughput of Coexistence Model with Multiple WPANs In this section, we derive the WLAN throughput of the coexistence model between a WLAN and multiple WPANs to evaluate the efficiency of the proposed synchronization method. For simplicity, the beacon intervals of multiple WPANs are the same value. WLAN throughput in coexistence with multiple WPANs (S Multi ) is given by the following equation, 5. Numerical Results 5.1 Coexistence Model with Representative H-STA Computer Simulation Results We evaluate the performance of superframe acquisition by the representative H-STA. We employ a Monte Carlo simulation platform using the C language to evaluate the proposed resource protection scheme for WPANs. Table 1 summarizes major parameters of the conventional ACROS scheme and the proposed HYSAC scheme. The transmission priority of the ACROS devices or H-STA is adjusted by its CWmin value. The superframe duration and the beacon interval are set to ms and ms, respectively. Figure 11 shows the simulation results for the WPAN protection failure probability versus the WPAN protection attempt period for ten WLAN STAs. As the WPAN protection attempt period increases, H-STA obtains more opportunities to transmit the RTS frame before WPAN beacon frame transmission; therefore, the WPAN protection failure probability decreases in the proposed scheme. An RTS frame is transmitted just once in the conventional ACROS scheme; therefore, the WPAN protection failure probability does not Table 1 Major parameters of the WLAN and WPAN. Parameters Values WLAN Data rate 24 Mbit/s MAC payload of AP, STA 1500 B, 300 B Offered load Saturated CWmin of AP, STA 3, 15 CWmax of AP, STA 255,1023 Retry limit 7 Slot time 9 μs SIFS 16 μs WLAN part CWmin 0, 1, 2 of H-STA CWmax 1023 and ACROS Retry limit WPAN Beacon interval ms Superframe duration ms S Multi = S total [1 (1 P PF (T PAP )) αd], (21) where α denotes an expected value of the number of NAV periods during a WPAN beacon interval. As shown in Fig. 3(b), H-STAs without the synchronization method may not set a NAV period. Therefore, α becomes lower value than the number of WPANs. α is derived from computer simulation. When the synchronization method is employed, α = 1. Because multiple WPANs transmit their beacons and obtain their superframes in the same NAV period. Fig. 11 Simulation results of WPAN protection failure probability versus WPAN protection attempt period for ten WLAN STAs.

8 INOUE et al.: HYBRID STATION AIDED COEXISTENCE SCHEME BETWEEN WIRELESS PANS AND WIRELESS LAN 585 Fig. 12 Simulation results of WLAN throughput versus WPAN protection attempt period for ten WLAN STAs. decrease. As a CWmin value of H-STA decreases, it obtains more opportunities to transmit the RTS frame than WLAN nodes during the WPAN protection attempt period; therefore, the WPAN protection failure probability decreases. Figure 12 shows the simulation results for the WLAN throughput versus the WPAN protection attempt period for ten WLAN STAs. As the WPAN protection attempt period increases, the period that WLAN STAs set the NAV duration increases. This causes a monotonic decrease in the WLAN throughput. The WPAN protection failure probability of the ACROS scheme is higher than that of the HYSAC scheme. The ACROS scheme fails to protect the WPAN superframe period more often than the HYSAC scheme does. Therefore, the WLAN throughput in the ACROS scheme is higher than that in the HYSAC scheme. As a CWmin value of H-STA decreases, the WPAN protection failure probability monotonically decreases. As a result, the WLAN throughput decreases. In the HYSAC schemes, there is a trade-off relationship between the WPAN protection failure probability and the WLAN throughput. Generally, the frame loss rate is required to be less than 10% including the frame detection error in the PHY layer. The WPAN protection failure probability is required to be sufficiently less than 10%. We set the target value of the WPAN protection failure probability to be 1%. As shown in Fig. 11, the ACROS scheme is unable to achieve this target value for ten WLAN STAs. On the other hand, the HYSAC scheme sufficiently achieves the target value of the WPAN protection failure probability. Figures 11 and 12 show that the WLAN throughput degradation in the HYSAC scheme is 9.5% when the WPAN protection attempt period of 8 ms and CWmin H-STA = Analytical Results Figure 13 shows the numerical results for the WPAN protection failure probability versus the WPAN protection attempt period in the HYSAC scheme for ten WLAN STAs using simulation and theoretical analysis. Both results show the same tendency. However, the difference between them is Fig. 13 Analytical results of WPAN protection failure probability versus WPAN protection attempt period in the HYSAC scheme for ten WLAN STAs. Fig. 14 Analytical results of per-slot data frame transmission probability by WLAN nodes after the first collision of RTS frame with data frame. (Number of STAs: 10, CWmin AP = 3, CWmin STA = 15) large when the CWmin value of H-STA is zero. One of the reasons for the difference is the approximation in Eq. (6). We assumed that the data frame transmission probability by WLAN nodes is a constant value τ regardless of the number of RTS frame retransmissions by the H-STA. However, the data frame transmission probability by WLAN nodes after the RTS frame collision becomes smaller compared with that of the steady state value. This is because WLAN nodes tend to have a large backoff counter value due to collisions between RTS frames and data frames. The collision probability of the RTS frame in the analysis is higher than simulation when the RTS frames are retransmitted. Hence, the WPAN protection failure probability of the analytical results becomes higher than that of simulation results. This difference is significant when the CWmin value of H-STA becomes small. This is because the probability distribution of the backoff countervalueof WLAN nodesreturn to that of steady state during the interval of RTS frame retransmissions when CWmin value of H-STA is large. To show that the above approximation is the main cause

9 586 Fig. 15 Modified analytical results of WPAN protection failure probability versus WPAN protection attempt period in the HYSAC scheme for ten WLAN STAs and CWmin H-STA = 0. Fig. 16 Analytical results of WLAN throughput versus WPAN protection attempt period in the HYSAC scheme for ten WLAN STAs. of the difference, we derive the per-slot data frame transmission probability by WLAN nodes according to the number of RTS frame retransmissions when the CWmin value of H- STA is zero. At the RTS frame retransmission, the probability distribution of the backoff stage and the backoff counter value of WLAN nodes varies from that of the steady state. For simplicity, we consider the variation of probability distribution of AP. This is because the CWmin value of AP is small to support QoS, therefore the variation of distribution of AP has a significant impact on the data frame transmission probability of WLAN nodes. Figure 14 shows analytical results of the per-slot data frame transmission probability by WLAN nodes at steady state and after the first collision of RTS frame of H-STA with data frame of AP or STAs. Slot index j = 0 represents just after DIFS. This result shows the AP has a significant impact on the decrease of the transmission probability when the RTS frame is retransmitted. We calculate the WPAN protection failure probability by using the modified probability distribution of AP and show the results in Fig. 15. The results of simulation and modified theoretical analysis are in good agreement. Those results show that the approximation of the data frame transmission probability is not appropriate when there are the WLAN nodes that have small CWmin value. Figure 16 shows the analytical results for the WLAN throughput versus the WPAN protection attempt period in the HYSAC scheme for ten WLAN STAs by using simulation and theoretical analysis. The difference between both results is less than 0.3%. 5.2 Coexistence Model with Multiple WPANs According to Sect. 5.1, the CWmin value of the H-STA and WPAN protection attempt period are set to zero and 8 ms, respectively so that WPAN protection failure probability becomes less than 1%. The other parameters are set to the same values as in Sect 5.1. We assume that the assignment of the representative H-STA and synchronization of the bea- Fig. 17 WLAN throughput versus the number of WLAN STAs (r: the number of WPANs). The HYSAC scheme without synchronization indicates that the H-STAs asynchronously transmit the beacon frames. con transmission are maintained by the proposed method and that the control information exchange and determination of the representative H-STA have been completed before the steady state communication. The HYSAC scheme without synchronization indicates that the H-STAs asynchronously transmit the beacon frames. Figure 17 shows simulation results of WLAN throughput performance versus the number of WLAN STAs when the number of WPANs is from one to four. Regardless the number of WLAN STAs, the throughput of the proposed method is higher than that of HYSAC scheme without synchronization. In the proposed method, up to four WPANs can operate in the same period. Thus, the WLAN throughput values for one to four WPANs are the same. In HYSAC scheme without synchronization, WLAN throughput decrease as the number of WPANs increases. WLAN throughput of the proposed method is 30.6% greater than that of HYSAC scheme without synchronization for ten WLAN STAs and four WPANs.

10 INOUE et al.: HYBRID STATION AIDED COEXISTENCE SCHEME BETWEEN WIRELESS PANS AND WIRELESS LAN Conclusion This paper has proposed a coexistence scheme, called HYSAC, between multiple IEEE based WPANs and an IEEE g based WLAN at the 2.4 GHz band. The HYSAC scheme employed an H-STA which consists of an IEEE based WLAN STA part and an IEEE based WPAN PAN-C part. We derived the performance of the proposed scheme theoretically when a QoS supported WLAN and a representative WPAN coexist. The analytical results and simulation results are in a good agreement. The proposed HYSAC scheme outperformed the conventional ACROS scheme with regard to the WPAN superframe period protection performance. The simulation results show that the WPAN protection failure probability of the conventional scheme and the proposed scheme are 31%, and less than 1%, respectively for ten WLAN STAs under saturated traffic conditions. Moreover, this paper has proposed the synchronization method for the instants of beacon transmissions between multiple WPANs. This method enables multiple WPANs to operate simultaneously. The numerical results showed that the WLAN throughput of the proposed method becomes 30.6% higher than that of HYSAC scheme without synchronization for ten WLAN STAs under saturated traffic conditions. References [1] IEEE Std , Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications, [2] IEEE Std , Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (LR-WPANs), [3] T.O. Kim, J.S. Park, H.J. Chong, K.J. Kim, and B.D. Choi, Performance analysis of IEEE non-beacon mode with the unslotted CSMA/CA, IEEE Commun. Lett., vol.12, no.4, pp , April [4] K.J. Myoung, S.Y. Shin, H.S. Park, and W.H. Kwon, IEEE b performance analysis in the presence of IEEE interference, IEICE Trans. Commun., vol.e90-b, no.1, pp , Jan [5] S.Y. Shin, H.S. Park, S. Choi, and W.H. Kwon, Packet error rate analysis of ZigBee under WLAN and Bluetooth interferences, IEEE Trans. Wirel. Commun., vol.6, no.8, pp , Aug [6] C. Won, J.H. Youn, H.A. Sharif, and J. Deogun, Adaptive radio channel allocation for supporting coexistence of and b, Proc. IEEE Vehiclar Technology Conference, pp , Dallas, Texas, Sept [7] N. Torabi, W.H. Kwon, and V.C.M. Leung, A robust coexistence scheme for IEEE wireless personal area networks, Proc. IEEE Consumer Commun. and Networking Conf., pp , Las Vegas, USA, Jan [8] K. Hwang, S. Yeo, and J.H. Park, Adaptive multi-channel utilization scheme for coexistence of IEEE LR-WPAN with other interfering systems, Proc. 11th IEEE Int. Conf. High Performance Computing and Commun., pp , Seoul, Korea, June [9] S.Y. Shin, D.H. Woo, J.W. Lee, H.S. Park, and W.H. Kwon, Active channel reservation for coexistence mechanism (ACROS) for IEEE and IEEE , IEICE Trans. Commun., vol.93, no.8, pp , Aug [10] C. Na, J.K. Chen, and T.S. Rappaport, Measured traffic statistics and throughput of IEEE b public wlan hotspots with three different applications, IEEE Trans. Wirel. Commun., vol.5, no.11, pp , Nov [11] I. Tinnirello and G. Bianchi, Rethinking the IEEE e EDCA performance modeling methodology, IEEE/ACM Trans. Netw., vol.18, no.2, pp , April [12] G. Bianchi, Performance analysis of the IEEE distributed coordination function, IEEE J. Sel. Areas Commun., vol.18, no.3, pp , March a member of IEEE. Fumihiro Inoue received his B.E. degree from Kyoto University, Japan, in He is currently an M.E. student at Kyoto University, Japan. He has been engaged in Protection Scheme of Wireless Sensor Network against Wireless LAN. He is a student member of IEICE. Takayuki Nishio received the B.E. degree in Electrical and Electronic Engineering from Kyoto University in 2010, and the M.I. and Ph.D. degree in Communications and Computer Engineering, Graduate School of Informatics from Kyoto University, Kyoto, Japan, in 2012 and 2013 respectively. In 2013, he joined the faculty of Communications and Computer Engineering, Graduate School of Informatics, Kyoto University, Japan, where he is currently an Assistant Professor. Masahiro Morikura received his B.E., M.E., and Ph.D. degrees in electronics engineering from Kyoto University, Kyoto, Japan in 1979, 1981, and 1991, respectively. He joined NTT in 1981, where he was engaged in the research and development of TDMA equipment for satellite communications. From 1997 to 2002, he was active in the standardization of the IEEE a based wireless LAN. Dr. Morikura is now a professor in the Graduate School of Informatics, Kyoto University. He is

11 588 Koji Yamamoto received the B.E. degree in electrical and electronic engineering from Kyoto University in 2002, and the M.E. and Ph.D. degrees in informatics from Kyoto University in 2004 and 2005, respectively. Since 2005, he has been with the Graduate School of Informatics, Kyoto University, where he is currently an associate professor. His research interests include game theory, spectrum sharing, and M2M networks. He is a member of the IEEE. Fusao Nuno received his B.E. degree from Ehime University in 1991 and M.E. degree from Kumamoto University in 1993, respectively in Electrical Engineering. In 1993, he joined NTT and engaged in research and development of portable terminals and base stations for broadband wireless access systems. He is currently a senior research engineer in NTT Access Network Service Systems Laboratories and engaged in the development of the wireless access systems for Wide Area Ubiquitous Network. Takatoshi Sugiyama received the B.E., M.E. and Ph.D. degrees from Keio University, Japan in 1987, 1989 and 1998, respectively. Since joining NTT in 1989, he had been engaged in the research of interference compensation, CDMA, MIMO, OFDM technologies for satellite, wireless LAN systems. He is currently a senior research engineer, supervisor, Group Leader in NTT Access Network Service Systems Laboratories. He received the Young Engineers Award from the IEICE of Japan in He is a member of the IEEE.