KR100983049B1 - Traffic-aware decentralized ap selection for multi-rate in wlans - Google Patents

Abstract

PURPOSE: A dispersed AP(Access Point) selection method through the traffic recognition in radio environment using multiple transmission rate based on a beacon frame is provided to broadcast a beacon frame, thereby selecting AP which can be expected the highest treatment amount among a plurality of APs neighboring to a client. CONSTITUTION: An AP measures a retry field value of incoming packet transmitted from each client within a BBS(Basic Service Set)(S410). The AP measures a goodput during constant time. The AP uses the retry field value and the goodput to calculate throughput/client per client, and idle probability(S420). The AP transmits a frame including the retryratio, the treatment amount per client, and a frame including the idle probability to at least one client within a BBS(S430).

Description

Decentralized AP Selection Method through Traffic Recognition in Wireless Environments Using Multiple Transmission Speeds

The present invention relates to a method for selecting an access point, and more particularly, a distributed access point selection method for recognizing traffic in a wireless environment using multiple transmission speeds and selecting an access point having excellent throughput among a plurality of access points. It is about.

With the recent popularization of IEEE 802.11 wireless LAN equipment, wireless equipment can be used anywhere, anytime. Wireless devices offer relatively satisfactory throughput with mobility and flexibility when compared to wired devices.

In order to use a service such as the Internet through a wireless LAN, each client must select an appropriate access point (AP). This choice of appropriate AP plays a critical role in the performance of the system, including throughput, fairness and quality of service, but the 802.11 standard does not provide detailed criteria for AP selection. have.

Typically, AP manufacturers use RSSI (Received Signal Strength Indication) as a method for each client to select an AP. That is, each client searches for an AP near its own to select an AP having the largest RSSI signal.

The AP selection method that considers only the channel quality like this causes a concentration of clients to receive a service from a specific AP, resulting in an imbalance in network resources.

On the other hand, the AP selection method using RSSI generally assumes that each station uses the same data rate. However, in order to cope with the variability of radio channels and to efficiently use space, the current 802.11 physical layer is 54Mbps in case of IEEE 802.11a / g. In case of IEEE 802.11b, it supports multi-rate up to 11Mbps. However, when there are stations with low data rates in such an environment, the performance of the system as a whole decreases considerably, which creates a situation in which a low data rate station occupies the channel for a long time, which is disadvantageous to a high data rate station. This is because the throughput of the station becomes a downward leveling which is equal to the throughput of the station having the lowest transmission rate. This phenomenon is called rate-anomaly.

It is also assumed that the neighboring APs use orthogonal channels that do not affect each other in an ideal state, but in reality the limit of radio channels without interference (three for 802.11b / g and 12 for 802.11a) ), Co-Channel Interference occurs. Transmission between adjacent BSSs, called Inter-BSS interference, can cause system performance degradation.

To solve this problem, M. Abububaih, et al., "A new access point selection policy for multi-rate IEEE 802.11 WLANs" (International Journal of Parallel, Emergent and Distributed Systems, vol. 23, no. 4, pp. 291-307, Aug. 2008 (hereinafter referred to as "IJPEDS") has proposed an AP selection method considering multiple data rates, but still does not consider transmission between adjacent BSSs.

Meanwhile, K. Sundaresan and K. Papagiannaki's "The need for cross-layer information in access point selection algorithms" (in proc ACM Sigcomm IMC, pp.257-262, Oct. 2006) (hereinafter "IMC") Both factors are taken into account, but the problem is that these factors are not accurately reflected.

The present invention is to solve the problems of the prior art, and provides a new AP selection method that can select the AP that can expect the highest throughput among a plurality of APs adjacent to the client in a wireless environment using a multi-transmission rate. I would like to.

An access point selection method for achieving the object of the present invention is a method for selecting an access point in a wireless environment using a multi-transmission rate, the access point is delivered from each client in a basic service set (BSS) A first step of measuring a value of a Retry field of an incoming packet, a second step of measuring goodput for a predetermined time determined by the access point, and a value of the Retry field measured by the access point and the goodput A third step of calculating RetryRatio, throughput per client, and idle probability using the second step; and transmitting, by the access point, a frame including the RetryRatio, throughput per client, and idle probability to at least one client in the BSS. It is made, including.

Here, in the fourth step, the access point broadcasts a beacon frame including the RetryRatio and the throughput per client and the idle probability, or the fourth step allows the access point to probe from at least one client in the BSS. Step 4-1 of receiving a request frame, and step 4-2 of the access point transmitting a probe response frame including the RetryRatio, the throughput per client, and the idle probability to the client that sent the probe request frame; It may include.

In another aspect of the present invention, an access point selection method includes a client receiving a frame from two or more access points, the frame including RetryRatio and throughput and idle probabilities for each client in the basic service set of each access point; Calculating a predicted throughput of the client when connecting to each of the access points using the RetryRatio and the throughput per client and the idle probability, and calculating the predicted throughput of the client when connecting to each of the calculated access points. And selecting one of the two or more access points based on that.

Meanwhile, the access point reselection method of the present invention includes the steps of a client receiving a current RetryRatio from a first access point connected to the client, and a previous RetryRatio previously received by the client from the first access point. And comparing the current RetryRatio and the previous RetryRatio as a result of the comparison, and determining a handover when the difference exceeds a predetermined threshold.

Wherein the client receives a frame from two or more second access points not associated with the client, the frame including RetryRatio and throughput and idle probabilities for each client in the basic service set of each second access point; Calculating a predicted throughput of the client when the client connects to each of the second access points using the RetryRatio, the throughput per client, and the idle probability; and when the client connects to each of the calculated second access points, And selecting one of the two or more second access points based on the predicted throughput of the client.

As described above, the present invention more accurately calculates the predicted throughput of each AP by estimating the number of active clients in each BSS using the Retry field in the MAC header, considering the multiple data rates and transmissions between neighboring BSSs. can do. Based on this predictive throughput, each client can choose the most appropriate AP. Therefore, each client can get the best throughput, and the performance of the whole system can be improved.

Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in various forms, and only the present embodiments are intended to complete the disclosure of the present invention, and the general knowledge in the art to which the present invention pertains. It is provided to fully convey the scope of the invention to those skilled in the art, and the present invention is defined only by the scope of the claims. Meanwhile, the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. In the present specification, the singular form includes plural forms unless otherwise specified in the specification. As used herein, “comprises” and / or “comprising” refers to the presence of one or more other components, steps, operations and / or elements. Or does not exclude additions.

The present invention proposes a new access point selection method that can select an optimal access point by estimating traffic on a wireless network.

1 is a diagram illustrating a wireless LAN environment using a multiple transmission rate to which an access point selection method according to an embodiment of the present invention is applied.

As shown in FIG. 1, a plurality of access points (APs) 210, 220, 230, and 240 exist around the user 100, and the user should select one of the APs. Meanwhile, in the present invention, it is assumed that the same channel state is shown to each node in one BSS (Basic Service Set), and all nodes in the BSS are within a range capable of wireless communication.

According to the IEEE 802.11 standard, the AP association procedure consists of passive discovery, active discovery, and association phase. The purpose of the search is to find a suitable AP that can provide the best performance among the available APs. There are two kinds of discovery (passive discovery, active discovery) in the wireless environment of infrastructure mode. In passive discovery, the AP periodically broadcasts a beacon frame containing BSS-specific information such as Timestamp and SSID. The client listens for beacon frames from each channel and selects the appropriate AP according to the signal strength. In the active search, each client broadcasts a probe request frame, and the AP that listens immediately sends a probe response frame to the client. After receiving several probe response frames, the client selects an appropriate AP according to the signal strength. After the discovery ends, the client exchanges a series of authentication and connection request / response frames and then establishes a connection with the target AP.

In the embodiment of the present invention, the throughput of the client is estimated by using the Retry field of the MAC header, and the access point that provides the best throughput is selected by using the same. In other words, the Retry field is used as channel feedback indicating channel status.

2 shows a configuration of a MAC header.

As shown in FIG. 2, there is a Retry field among several fields configuring the first Frame Control field of the MAC header. The Retry field consists of 1 bit and has a value of 0 when the first transmission attempt is successful, and a value of 1 when the frame is retransmitted due to failure at the first transmission attempt.

Thus, this field can be used to estimate the frequency of retransmissions to estimate the number of clients currently active.

라 하 면, 예측 처리량은 다음의 [수학식 1]과 같다. The predicted throughput of each client can be defined as the inverse of the time taken to successfully transmit one data frame. In other words, the average time until client k successfully transmits data to AP i

In this case, the predicted throughput is given by Equation 1 below.

In Equation 1, retransmission due to collision between stations is not considered. Accordingly, in consideration of this point, the average time taken until the client k successfully transmits the data frame of length L to the AP i after j times due to a collision can be expressed as Equation 2 below.

Where rate (k) is the data rate of client k and tDATA is determined by the data rate of client k. l _macheader , l _payload and l _ack mean the length of MAC header, length of payload and length of ACK frame, respectively. Control frames such as ACK, Request to Send (RTS), and Clear to Send (CTS) are transmitted at the basic rate according to the 802.11 standard. For example, for 802.11b, the rate _basic is 1 Mbps. tACK means the duration of the ACK frame, tDIFS means the duration of the DIFS, tSIFS means the duration of the SIFS.

In addition, the average backoff time (tbackoff (j)) during successive attempts to transmit j times may be expressed as in Equation 3 below.

Here, the contention window initially takes the CW _min value, and each time the data transmission attempt fails, the CW value is doubled until it reaches CW _max . If the CW value reaches CW _max , the CW value remains at the CW _max value until the data transmission is successful or the data frame is discarded because of the Retry limit.

Here, the operation of the backoff counter should be considered when calculating tbackoff (j). If the channel is congested during the backoff slot, the backoff process is stopped. Suppose a station's backoff counter is b. If the current channel is idle, the backoff counter is decremented by one at the end of slot time, so the backoff counter is b-1 at the next slot time. On the other hand, if the current channel is congested, the station's backoff counter remains at b. That is, the backoff counter decrements only when the channel is idle. After all, when calculating tbackoff (j), slot time represents virtual slot time, not absolute (physical) slot time.

E [slot time] represents the average length of slot time, which is the Bianchi model (G. Bianchi, "Performance Analysis of the IEEE 802.11 Distributed Coordination Function", IEEE J. Select.Areas Commun, vol. 18, no. 3 , pp. 535-547, Mar. 2000) (hereinafter "Bianchi") has the same meaning as the virtual slot time, and is represented by Equation 4 below.

Where T represents the time measured to estimate the average length of the slot time. P _idle is the idle probability of idleness during T time, and P _tr is the probability that at least one station transmits a packet at any point in time. Pc is a conditional collision probability, and means a collision probability when packet transmission occurs on a channel, which is the same as p to be described later. 1-Pc is a conditional probability, which is the probability that the transmission on the channel will succeed. That is, the probability that only one station transmits on the channel under the condition that at least one station transmits. σ is the period of empty slot time, Ts is the average time that the channel feels congested due to successful transmission, and Tc is the average time that the channel feels congested due to collision.

Ts and Tc are determined by the lowest transmission rate among clients already connected to AP i. Therefore, estimating P _idle plays an important role in deriving E [slot time]. Assuming that each station in the BSS sees the same channel view, P _idles of all stations in the same BSS have the same value. Therefore, each AP can estimate the P _idle of its BSS by measuring the channel state.

Therefore, the average time required for client k to successfully transmit data is expressed by the following equation (5).

은 m번째 전송시도가 실패되었을 때 소요되는 시간을 뜻하고, 아래의 [수학식 6]과 같다.Here, γ is the Retry Limit, p is the average uplink transmission failure probability of each BSS, the derivation process will be described later.

Is the time taken when the m-th transmission attempt fails, and is shown in Equation 6 below.

However, Equation 5 does not consider rate-anomaly, another problem in a multi-rate wireless LAN environment. That is, if a new client selects an AP that is already in use by a client using a lower transmission rate than its own, the predicted throughput of the new client is leveled down to the throughput of the slowest client by rate-anomaly.

In order to solve this problem, the AP selection method of the present invention uses the goodput of the AP for a predetermined time (for example, 5 seconds). If the goodput of the AP for a certain time is S, the throughput per client in each BSS _i for a long time can be expressed as Equation 7 below.

Therefore, the estimated throughput of the client considering the rate-anomaly can be expressed as Equation 8 below.

As mentioned above, in the AP selection method of the present invention, the frequency of retransmission is measured to estimate the number of currently active clients or the probability of transmission failure. The frequency of retransmission is measured using the Retry field in the header of the MAC layer frame. The Retry field is 1 bit in length and indicates whether data or a management frame is first transmitted or retransmitted.

In other words, if the Retry field is set to 0, the frame is the first one transmitted, and if it is set to 1, the frame is a retransmission for a previously failed frame. This field is used to remove duplicate frames by the receiver, but the Retry field can be used as feedback to the channel to estimate the state of the channel.

The more congested the channel, the higher the frequency of retransmissions. Using this point, the receiving side can predict the average uplink transmission probability (p) of each BSS. If the number of Retry fields is set to 1 during the measurement time, the transmission attempt is frequently failed due to collision. Because you can see.

To analyze the pattern of the Retry field, Bianchi's Markov chain model (“Bianchi”), which represents the backoff window scheme of the 802.11 DCF, is shown as shown in FIG. 2. When b (t) represents the backoff window size and s (t) represents the backoff stage, the state distribution of the chain is represented by Equation 9 below.

As shown in FIG. 3, transmission is performed when the backoff time counter is zero. The Retry field is set to 0 when transmission is made in backoff stage 0 (when shown in bold solid line in FIG. 3), and the Retry field is set to 1 when transmission is made in another stage (shown in dashed line in FIG. 3). Occation). Therefore, the probability of successful transmission in the first attempt can be calculated as shown in Equation 10 below.

Here, i represents a value in which the Retry field is set, and thus i is 0,1 and Ci is the number of frames in which the value of the Retry field is i. Then C ₁ / C ₀ can be called RetryRatio. m is the maximum value of the backoff stage.

Number of stations Transmission failure probability RetryRatio Number of stations Transmission failure probability RetryRatio One 0.000 0.000 11 0.308 0.441 2 0.059 0.062 12 0.322 0.470 3 0.107 0.120 13 0.335 0.497 4 0.147 0.173 14 0.346 0.522 5 0.181 0.221 15 0.357 0.547 6 0.210 0.265 16 0.402 0.654 7 0.235 0.306 17 0.436 0.745 8 0.256 0.343 18 0.463 0.824 9 0.276 0.378 19 0.507 0.960 10 0.293 0.411 20 0.540 1.075

Now, using these results, we explain how the client selects an AP.

According to the embodiment of the present invention, after performing a passive / active search, the client should estimate the uplink transmission failure probability p of each BSS. However, a newly arrived client cannot derive the correct p value because of hidden nodes. At this time, if the AP periodically includes a RetryRatio directly related to p in a beacon frame and broadcasts it or transmits it in a probe response frame to the client, each client can derive an accurate p value by listening to these frames. Will be.

RetryRatio is a good criteria for selecting AP because the traffic trend for a certain time (for example, 5 seconds) is reflected based on the time when the beacon / probe response frame is sent, and the RetryRatio included in the beacon / probe response frame The size of the field is only a few bytes, so it is not a big overhead.

In addition, if each AP includes P _idle in the beacon / probe response frame and transmits them periodically, each client can obtain the E [slot time] for each BSS by listening to the frames.

4 is a flowchart illustrating a process in which each AP transmits a beacon frame including such information.

As shown in Figure 4, each AP measures the value of the Retry field of the incoming packet, and measures goodput (Si) for a predetermined time (S410).

Next, each AP calculates {RetryRatio, S, P _idle } using the measured value (S420), and transmits (broadcasts) a beacon frame including the same (S430).

Then, each client can obtain the predicted throughput of each client when connecting to each AP by using {RetryRatio, S, P _idle } information in the received beacon frame, and selects an appropriate AP accordingly.

The method described with reference to the flowchart of FIG. 4 represents a case of passive discovery, and in the case of active discovery, when a client broadcasts a probe request frame to each access point, each access point transmits {RetryRatio, S, By transmitting a probe response frame including P _idle } information, each client can be provided with information for selecting an appropriate AP.

In addition, the AP selection method according to an embodiment of the present invention may be applied to a handoff process reflecting dynamic traffic.

That is, each client independently selects an appropriate AP according to the above-described AP selection method, but the selected AP is not the optimal AP over time due to the change in the number of clients as the number of new clients increases or irregularity of traffic. Can be.

Therefore, in order to cope with such changes in the wireless environment, a dynamic AP selection method can be used. This dynamic AP selection method is shown in FIG.

As shown in FIG. 5, each client receives a RetryRatio from each AP (S510). However, if the AP receiving the RetryRatio is the currently connected AP and the newly received RetryRatio differs from the previously received RetryRatio by more than a predetermined threshold (for example, 0.3) (S520), the AP is considered to be an appropriately selected AP. It becomes impossible situation. Therefore, in this case, the handover is determined (S540), and reconnection with another AP is attempted (S560). This method is called dynamic AP selection. When attempting to reconnect with another AP, the AP selection method as described above may be used.

That is, when the AP that transmits the RetryRatio is not the connected AP (S530), the expected throughput of the corresponding AP is calculated (S550), and used as information for reconnection.

Simulation was performed to test the performance of the AP selection method according to an embodiment of the present invention, the results are shown in FIG.

We simulated the ns-2 (version 2.33) simulator and modified the beacons and probe frames in 802.11 DCF mode. The 802.11b physical layer is used, and the carrier sensing distance is set to 550m. The transmission rate of each client is determined only by the distance between the client and the AP, and the transmission distance according to the transmission rate is referred to the specification of the OriNOCO 11b Card.

Path loss of the radio signal depends on the TwoRayground model implemented in ns-2. All clients and APs use the same transmission power. Considering a multi-cell environment in which nine APs use the same channel, all clients are arbitrarily arranged to connect to at least one AP. The reason for selecting such a simulation environment is to prove the effectiveness of the AP selection method of the present invention in an environment in which transmission between adjacent BSSs exists. Without rate adaptation, client arrival times are evenly distributed within the first 30 seconds of the simulation. During the simulation, each client generates upstream traffic to its AP, where the traffic provided is CBR UDP traffic and the packet size is 1000 bytes. It is also assumed that every client in the system always has a packet to send to the AP. At this time, both the AP selection method of the present invention and the IJPEDS and IMC methods were simulated by the static AP selection method without the re-association process.

The simulation was performed for 100 seconds, and the aggregated throughput measurements were made for a total of 68 seconds from 32 seconds to the rest of the time. Thus the AP selection method of the present invention was compared with the method of IJPEDS and IMC. In order to simulate the AP selection method proposed by IJPEDS and IMC, Intersil's Bit Error Rate (BER) vs. Signal-to-noise curves are used. All results are carried out 10 times and the average value is taken.

FIG. 6 is a graph illustrating aggregated throughput of an AP selection method and an existing approach according to an embodiment of the present invention whenever the number of clients increases from 50 to 80. FIG. While the number of clients increases, the aggregate throughput of all AP selection methods increases linearly until the saturation time point. After this point, even though the number of clients increases, the aggregate throughput no longer increases, but decreases drastically. As shown in Figure 6 it can be seen that the AP selection method according to an embodiment of the present invention shows superior performance compared to other approaches. That is, the AP selection method according to the embodiment of the present invention is superior to 11.3% performance compared to IJPEDS, and 12% better than IMC.

The reason for this difference in performance is as follows. IJPEDS's approach does not reflect transmissions between neighboring BSSs, so there is a performance difference, and IMC's approach shows a large variation in the ATD measurement, which is directly related to the number of active clients. Affects the expected throughput, resulting in a wrong AP selection, resulting in performance differences.

Next, in the AP selection method according to an embodiment of the present invention, the performance change according to the presence or absence of a re-association procedure, that is, the handover function is simulated. Except for using the Exponential On / Off model to provide traffic variability, the rest of the parameters were simulated in the same manner as described above.

7 is a graph showing aggregated throughput with or without a re-association procedure of an AP selection method according to an embodiment of the present invention whenever the number of clients increases from 50 to 80. While the number of clients increases, the aggregate throughput of all AP selection methods increases linearly until the saturation time point. After this point, even though the number of clients increases, the aggregate throughput no longer increases, but decreases drastically. As shown in FIG. 7, it can be seen that a method having a re-association function, that is, a handover function, among the AP selection methods according to an embodiment of the present invention shows better performance than the case where the function is not implemented. That is, the AP selection method according to the embodiment of the present invention is superior in performance by 4.8% compared to the method without the handover procedure.

The AP selection method according to an embodiment of the present invention may be implemented as computer readable codes on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, hard disk, floppy disk, flash memory, optical data storage device, and also carrier waves (for example, transmission over the Internet). It is also included to be implemented in the form of. The computer readable recording medium can also be distributed over computer systems connected over a computer network so that the computer readable code is stored and executed in a distributed fashion.

Although the present invention has been described above with reference to preferred embodiments, the AP selection method of the present invention is not necessarily limited to the above-described embodiments, and various modifications and variations can be made without departing from the spirit and scope of the present invention. The appended claims will cover such modifications and variations as long as they fall within the spirit of the invention.

1 is a diagram illustrating a wireless LAN environment to which an AP selection method according to an embodiment of the present invention is applied;

2 is a diagram illustrating a configuration of a MAC header;

3 is a diagram showing a Markov chain model of the 802.11 DCF BEB,

4 is a flowchart illustrating an AP selection method according to an embodiment of the present invention from the perspective of an AP;

5 is a flowchart illustrating a dynamic AP selection method applying the AP selection method to a dynamic environment according to an embodiment of the present invention;

FIG. 6 is a graph illustrating a simulation result for static AP selection and aggregated throughput in the prior art according to an embodiment of the present invention; FIG.

FIG. 7 is a graph illustrating a simulation result for aggregated throughput according to whether or not a re-association procedure is used in an AP selection method according to an embodiment of the present invention.

Claims (6)

As a method of selecting an access point in a wireless environment using multiple transmission speeds, A first step of the access point measuring a value of a Retry field of an incoming packet delivered from each client in a basic service set (BSS), A second step of measuring goodput for a predetermined time determined by the access point; A third step of calculating a RetryRatio, a throughput per client, and an idle probability using the value of the Retry field measured by the access point and the goodput; And a fourth step, wherein the access point transmits a frame including the RetryRatio, the throughput per client, and the idle probability to at least one client in the BSS. The method of claim 1, wherein in the fourth step, And the access point broadcasts a beacon frame including the RetryRatio, the throughput per client, and an idle probability. The method of claim 1, wherein the fourth step, Step 4-1 in which the access point receives a probe request frame from at least one client in the BSS; And accessing, by the access point, a probe response frame including the RetryRatio, the throughput per client, and an idle probability to the client that sent the probe request frame. As a method of selecting an access point in a wireless environment using multiple transmission speeds, Receiving, by a client, a frame comprising RetryRatio and throughput and idle probabilities for each client in the basic service set of each access point from two or more access points; Calculating a predicted throughput of the client when the client connects to each of the access points using the RetryRatio, the throughput per client, and the idle probability; Selecting one of the two or more access points based on the estimated throughput of the client when the client connects to each of the calculated access points. As a method of reselecting an access point in a wireless environment using multiple transmission speeds, The client receiving a current RetryRatio from a first access point associated with it, Comparing the current RetryRatio with a previous RetryRatio previously received by the client from the first access point; Determining a handover when a difference between the current RetryRatio and the previous RetryRatio exceeds a predetermined threshold as a result of the comparison; If the handover is determined, the client receives a frame including RetryRatio and throughput per idle client and idle probability in each of the basic service sets of each of the second access points from two or more second access points not connected to the client. Steps, Calculating a predicted throughput of the client when the client connects to each of the second access points using the RetryRatio, the throughput per client, and the idle probability; Selecting one of the at least two second access points based on the predicted throughput of the client when the client connects to each of the calculated second access points. delete

Images