WO2009093799A1

WO2009093799A1 - Methods and system for application-aware and power-efficient fast handover in wireless lan

Info

Publication number: WO2009093799A1
Application number: PCT/KR2008/006755
Authority: WO
Inventors: Jong Tae Park
Original assignee: Kyungpook National University Industry-Academic Cooperation Foundation
Priority date: 2008-01-23
Filing date: 2008-11-17
Publication date: 2009-07-30
Also published as: KR20090081292A; KR100949163B1

Abstract

The present invention relates to a layer-2 handover method and system in wireless local area network (WLAN), which can provide fast handover with guaranteed quality of service for applications (QoS) while minimizing the power consumption due to the handover operation. In order to do that, the scanning period and the number of scanned channels during the movement of a mobile node are adaptively adjusted in accordance with the type of the application and the WLAN conditions, so that the possible QoS degradation of application during handover is prevented while minimizing the power consumption. The present invention is particularly applicable to the complex wireless LAN environments such as indoor and metropolitan area in which the access points (APs) are densely installed.

Description

METHODS AND SYSTEM FOR APPLICATION- AWARE AND POWER-EFFICIENT FAST HANDOVER IN WIRELESS LAN

Technical Field

[1] The conventional techniques related to the present invention are described below.

Recently, there has been increasing interest in providing seamless connectivity for the multimedia application in wireless local area network (WLAN) while the users are moving around. Currently, the IEEE 802.11 a/b/g/n international standards of WLAN do not specify the methods for seamless handover between access points (APs) of WLAN. Accordingly, as the users are moving around different APs, the excessive amount of handover delay may occur. This may in turn result in disruption of real-time multimedia services such as VoIP and video conferencing during handover.

[2]

[3]

Background Art

[4] In Wireless LAN, the handover procedure consists of the four steps: channel discovery, channel switching, authentication, and association. Among these steps, the channel discovery is known to be a dominant factor for causing the handover latency, taking up almost 90 % time of the handover procedure. There have been diverse efforts to reduce the handover latency of channel discovery. Most of these research attempts to reduce the channel access time by gathering in advance the information of next APs which might be accessible during mobile node's movement. One representative method of these attempts is that of IEEE 802. I l k standard which utilizes the Neighbor Report for selective scanning. The Neighbor Report contains the information of access points (AP) s which are accessible from the current AP. In the selective scanning method, the channels in which APs are accessible are only scanned, and the optimal AP which provides the best performance with respect to the signal strength (and loading condition) is selected, so that the scanning time can be reduced. However, since this requires the modification of the APs, it may be difficult to widely implement this technique in the existing wireless LAN environment. Furthermore, this technique may not be suitable for the indoor and/or metropolitan area in which APs are densely installed.

[5] Recently, there appeared other fast layer 2 handover techniques which do not require the modification of the existing IEEE 802.11 standard. These include Proactive Scanning and Smooth Scanning which had been developed by Micro Soft Inc., and University of Massachusetts, respectively. Both of these methods employ the pre- scanning technique in which the neighboring AP information is gathered in advance during movement, and the APs are rapidly switched when the handover condition is satisfied. These methods however may incur the unnecessary large power consumption for handover preparation, and also non-negligible packet loss. Moreover, these methods do not reflect the QoS constraint of the streaming multimedia application during handover, In summary, most previous work on layer 2 handover mechansim do not reflect the varying QoS constraint of the multimedia application and wireless access network conditions, so that the quality of the application service during handover preparation and execution can be severely degraded, and the excessive additional power can be consumed for the handover operation.

[6]

[7]

Disclosure of Invention Technical Problem

[8] Accordingly, the present invention has been made keeping in an effort to solve the above problems occurring in the prior art, and it is an object of the present invention to provide fast handover methods and system which enable the quality of multimedia application service to be not degraded during handover at IEEE 802.11 a/b/g/n WLAN with minimized power consumption.

[9] Another object of the present invention is to make the scanning period and the number of scanned channels to be adaptively adjusted in accordance with the type of application and network condition in such a way that the degradation of service quality of the application service is prevented and the power consumption is minimized.

[10] Still another object of the present invention is to suggest fast handover methods and system which can be specially useful to the complex wireless network environments such as indoors and/or metropolitan area in which the APs are densely installed.

[H] [12]

[13]

Technical Solution

[14] The above technical problems can be solved by application- ware and power-efficient fast handover methods and system, which include the following steps of: gathering of neighboring AP information, handover preparation, and handover execution. Two threshold values, i.e., handover prepare threshold Tp and handover execute threshold Ts are defined to distinguish the above three steps. Tp and Ts are signal strengths which are determined by either received signal strength indicator (RSSI) or signal to noise ratio (SNR). [15] During the step of gathering of neighboring AP information, if the RSSI from the current AP is larger than Tp, the AP scanning period in pre- scanning operation is adjusted dynamically and the neighboring AP information is periodically gathered in a manner that the quality of application service is not degraded and the power consumption due to scanning operation is minimized. In this case, the number of channels which are accessed per one scan operation is also adjusted to minimize the power consumption, while making the service quality of application service to be not degraded.

[16] As the mobile node moves, the RSSI from the current AP may be less than or equal to Tp, and the mobile node enters into the handover preparation step. In handover preparation step, the candidate APs for the next AP are determined, and pre-scanning operation is reinitiated to these candidate APs with different scanning period. The candidate APs are selected, which show good performance in RSSI and AP/channel load condition during the previous step of gathering of neighboring AP information. The number of channels to be scanned is also adjusted by taking into account the type of application and network conditions. The scanning period is adjusted to be faster in order to reduce the handover latency, which in turn makes the candidate AP information to be obtained with greater accuracy.

[17] Lastly, as the RSSI becomes less than or equal to Ts due to the movement of the mobile node, it enters into the handover execution step., in which the AP with largest value of RSSI and with best performance in AP/channel load condition is selected as the next AP from candidate APs, and the handover operation is completed by changing the point of attachment from the current AP to the next AP.

[18]

[19]

Advantageous Effects

[20] The present invention has an advantage in that it can provide stable, power-efficient and fast handover to support QoS-guaranteed real-time mobile multimedia communication service such as mobile VoIP at indoor or metropolitan area with harsh electric wave characteristics.

[21] The present invention has another advantage that it can reduce a handover latency that may occur when a mobile node moves between neighboring cells, by implementing an application- aware and power-efficient handover method and system which enable the AP scanning period in pre-scanning operation to be adjusted dynamically, and the neighboring AP information is periodically gathered in a manner that the quality of application service is not degraded and the power consumption due to scanning operation is minimized during handover.

[22] The present invention has another advantage in that it enables handover stably and intelligently at harsh wireless network environment with fluctuating electric wave characteristics in a manner that the estimated average values of signal strength is used for determining the handover steps, rather than using signal strength. This results in effectively reducing the possible error in handover operation in complicated network conditions of indoors and/or metropolitan area.

[23] Moreover, when selecting the candidate AP, not only using the RSSI from the AP but also the AP/load condition is used, so that the quality of application service after handover can be maintained.

[24]

Brief Description of the Drawings

[25] Further objects and advantages of the invention can be more fully understood from the following detailed description taken in conjunction with the accompanying drawings in which:

[26]

[27] FIG. 1 shows variations of RSSI as the mobile node moves;

[28] FIG. 2 shows application buffer structure for multimedia streaming service;

[29] FIG. 3 is algorithm for Application Type Classification;

[30] FIG. 4 is application- aware and power-efficient handover algorithm;

[31] FIG. 5 shows structure of Neighbor_AP_List and Candidate_AP_List;

[32] FIG. 6 shows an example of application- aware and power-efficient handover algorithm;

[33]

Best Mode for Carrying Out the Invention

[34] The method includes the steps of a) preparing handover by pre-scanning operation with variable period and scan granularity which are adaptively changed by the type of application and the network condition in such a way that the quality of the application service is not degraded while the power consumption due to the pre-canning operation is minimized, b) if the estimate of the average signal strength from the current AP is less than or equal to Tp, then reducing the pre-scanning period and scan granularity in order to find out the optimal next AP by using both the signal strength and AP/load performance, and c) if the signal strength from the current AP is less than or equal to Ts, then changing the point of attachment from the current AP to the next AP, thereby completing the handover process. These steps have been designed in order to save power consumption and prevent data loss and delay during handover. These steps are distinguished by two threshold values, Tp and Ts, where Tp is the handover prepare threshold value and Ts is the handover execute threshold value. During the handover preparation, either the signal strength or the estimate of the average signal strength from the current AP may be used to compare with Tp, and during the handover execution step, not only the signal strength from the current AP but also the AP/load condition are used to select the best next AP. The present invention is particularly applicable to the complex wireless LAN environments such as indoor and metropolitan area in which the access points (APs) are densely installed.

[35]

[36]

Mode for the Invention

[37] The present invention will now be described in connection with specific embodiments with reference to the accompanying drawings. Fig. 1 shows the variations of RSSI, Handover Prepare Threshold, Tp, and Handover Execute Threshold, Ts, as the mobile nodes from the current AP to the next AP. In the current patent, the access point (AP) to which the mobile node currently is attached is called the current AP, and the access point to which the mobile node will be attached after handover is called the next AP. In the proposed application-aware algorithm, as the mobile node moves, if the RSSI value measured from the current drops to below Tp, the mobile node prepares for the handover in advance. If the RSSI value from the current AP becomes less than Ts and that from the next AP is beyond Ts, the mobile node rapidly changes the point of attachment from the current AP to the next AP, and completing the handover operation.

[38] The mobile node periodically transmits the probe message to the near-by APs in order to get the neighboring AP information. The response message form the AP includes the SSID and media access control (MAC) address of the AP, signal strength, and other information related to the authentication. The mobile node also gets the RSSI value from the current AP. Sometimes, signal to noise ration (SNR) values is used instead of RSSI. As shown in Fig. 1, if the RSSI from the current AP is beyond Ts, the mobile node transmits the probe message in order to get the near-by AP information. The mobile node can also get the near-by AP information by listening becon messages from the near-by APs. Since the wireless signal is very sensitive to the surrounding physical environment, the RSSI values usually fluctuate over time. In order to compensate for this unsteadiness of the wage signal strength, we can use the moving average value instead of RSSI. Let RSSI₁(AP) denote the value of RSSI which is received from the AP at time slot i. Let S₁(AP) denotes an estimate of the average RSSI which is received from the AP at time slot i. Then^AP) = (1 -

M

) S₁-I (AP) + [RSSI₁(AP) - RSSI₁^(AP)) . In other words, S₁IS an estimate of the moving average of radio signal strength. Generally,

is assigned the value of 0.4. In this way, the mobile node can get the near-by AP information more stably in a robust way, making the mobile node immune from abrupt fluctuation of radio signals in indoor and/or complex metropolitan area.

[39] Now, we describe the basic idea for the proposed application- aware and power- efficient handover mechanism. In WLAN, the service quality of the streaming service can be acceptable although the layer 2 connection is intermittently disconnected due to the temporary channel switching of pre-scanning phase. This is because the application buffer of the multimedia streaming service can store enough data to sustain the continuous delivery of the multimedia traffic during discontinuity of the layer 2. However, if the pre-scanning interval is too large, the service quality may be degraded, although the power consumption due to the pre-scanning operation can be minimized. If the pre-scanning interval is too short, the service quality may be again degraded with larger power consumption. The basic idea is to find out the optimal point of adjustment between pre-scanning interval and power consumption so that the pre-scanning operation to find out the neighboring AP information can be performed without any degradation of the service quality of the multimedia application, while minimizing the power consumption.

[40] Fig. 2 depicts the buffering structures at mobile node for multimedia streaming application. The mobile node starts to play out the streaming data after some amount of data is pre-fetched in the buffer as shown in Fig. 2. After playing out the service, the data is continuously entering into the buffer from the network, with rate of F(t), so the service can be provided without disruption. Here, F(t) is called Fill Rate, and it indicates the amount of data which arrives at the application buffer during unit time. In Fig. 2, D is the amount of data which leaves from the buffer during time unit. F(t) is termed as input rate, and D is called output rate. It is noted that F(t) varies along with time, being dependent of the network conditions, delay and bandwidth, but D is fixed and independent of time.

[41] We present the application-aware handover model and theory in which the QoS- guaranteed handover is achieved without service degradation of multimedia streaming service with power efficiency.

[42]

[43] Definition 1: The interrupt allowance time (AIAT) of an application service is defined as the time during which the streaming application can be played out without service degradation although the layer connection is disrupted due to scanning operation of the handover procedure. [44]

[45] Let us define the amount of pre-fetched data at time t as Q(t). Q(t) depends on the input rate F(t). For real-time streaming service, Q(t) can be obtained by estimating the moving average of F(t). D is usually constant, and depends on the type of codec. For MPEG 1 codec, D is 1.5 Mbps, and for MPEG 3, it is known to have the values between 3 Mbps and 6 Mbps. In general, the application interrupt allowance time depends on the buffer size, multimedia codec, and error concealment methods, and processing capability of the mobile node. As shown in Fig. 2, the multimedia application using streaming protocol is played out after the amount of data filled up at the application buffer reaches some threshold value Q. Therefore, even though the connections at the layer 2 is disrupted, the application can be continuously played out without service disruption until the pre-fetched data, i.e., Q bytes of data, in the application buffer is drained out. Therefore, the maximum of application interrupt allowance time for a multimedia application using streaming protocol is Q/D sec. Let us denote this Q/D time as T.

[46] For the stored video/audio service using the RTSP streaming protocol, the value of T is about between 2 sec and 5 sec. For VoIP application service, since the end-to-end delay should be less than 400 msec, T should be around 50 ms, taking into account the network condition, the type of VoIP codec and packet size of 20 ms. For non-real time applications such as file transfer and e-mail, the constraint of the value of T is not severe. The maximum value of AIAT may also depend on the performances of CPU and codec. Fig. 3 shows the algorithm for application type classification. In Fig. 3, the type of application can be identified by looking into the TCP/IP port number of the application. For the non-standard P2P application, it may be difficult to classify the type of the application, which is beyond the research scope of the present invention.

[47]

[48] Lemma 1: In wireless LAN, the maximum value of AIAT of an application at a mobile node is T.

[49] Proof: For a given time t, there exists Q(t) data at the application buffer. Since the streaming application can be at most played out without service degradation during Q(t)/D, T is the maximum value of AIAT. D

[50]

[51] Generally, the active scanning operation to get the neighbor AP information during movement incurs both packet loss and power consumption. This is because the current connection is temporarily disrupted during scanning operation, and the probing process in the active scanning requires power consumption which can otherwise be used for data packet transmission. Generally, the mobile node has the limited power, so that the power consumption for scanning operation should be minimized. [52]

[53] Lemma 2: As a mobile node moves in a wireless LAN, the handover procedure with minimal power consumption while sustaining the service quality of an application is performing one scan operation during T sec.

[54] Proof: By Lemma 1, an application at a mobile node can at most play out the service without service degradation during T sec, regardless to the number of scan operations of the layer 2 handover procedure. Since the power consumption for handover procedure is proportional to the number of scan operation, performing one scan operation during T sec results in the minimal power consumption for handover procedure while sustaining the service quality of the application. D

[55]

[56] Theorem 1: As a mobile node moves in a wireless LAN, the handover procedure for getting the best neighboring AP information with minimal power consumption while sustaining the service quality of an application is performing one scan operation during T sec.

[57] Proof: According to Lemma 2, the handover procedure with minimal power consumption, while sustaining the service quality of an application, is performing one scan operation during T sec. As a mobile node moves in a wireless LAN, the mobile node may get better information by performing more frequently the scan operation. However, these frequent scan operations increases the power consumption. Therefore, the scan operation for getting the best AP information with minimal power consumption is performing one scan operation during T sec. D

[58]

[59] Up to now, we have found that the optimum handover procedure with minimum power consumption while preserving the service quality of an application is performing one scan operation during the maximum value of interrupt allowance time of a multimedia application. However, the service quality of the multimedia application may degrade severely after handover procedure completes. This is because the packet loss may occur during the layer 2 handover operation, and the application buffer may not be filled in. In order to solve this problem, it is necessary to minimize the packet loss due to scan operation of the handover procedure. Generally, the packet loss which might have been generated during the layer 2 handover procedure may be recovered by various packet loss recovery mechanisms such as TCP error recovery, frame interleaving at application layer, and so on. However, TCP error recovery may not be suitable for time-critical real time application service, and frame interleaving technique may not be good enough for the high-quality real-time video conferencing service. Therefore, it is necessary to minimize the layer 2 packet loss due to handover procedure to guarantee the service quality of an application. [60]

[61] Lemma 3: Accessing one wireless channel during each scan operation results in minimum packet loss for the layer 2 handover procedure.

[62] Proof: Since the packet loss due to the layer 2 handover procedure is proportional the time to access a channel of the wireless LAN, it is also proportional to the number of wireless channels which is accessed during one scan operation. Therefore, accessing one wireless channel during each scan operation results in minimum packet loss. D

[63]

[64] Theorem 2: As a mobile node moves in a wireless LAN, the handover procedure with both minimal power consumption and minimal packet loss while sustaining the service quality of an application is probing one wireless channel during T sec.

[65] Proof: According to Theorem 1, the handover procedure with minimal power consumption while sustaining the service quality of an application is performing one scan operation during T sec. By Lemma 3, accessing one wireless channel during each scan operation results in minimum packet loss for the layer 2 handover procedure. Therefore, the handover procedure with both minimal power consumption and minimal packet loss while sustaining the service quality of an application is probing one wireless channel during T sec. D

[66]

[67] Corollary 2: Suppose that the application buffer can be filled up for scan operation during T sec during the movement of a mobile node at wireless LAN. In this case, if a wireless channel is accessed during scan operation during T sec, both power consumption and packet loss due to the handover procedure are minimized while preserving the service quality of the application.

[68] Proof: According to Theorem 2, we know that both minimal power consumption and minimal packet loss can be obtained by performing one channel access during T sec, while sustaining the service quality of an application. Since the application buffer is assumed to be filled up for each scan operation during T sec, the amount of data in the application buffer is larger than or equal to Q, i.e., T*D. Therefore, the service quality can be preserved for every scan operations. D

[69]

[70] In reality, the assumption in Corollary 2 may often not be satisfied due to the network congestion, fault, malfunction due to virus, etc. Furthermore, accessing one wireless channel during scan operation can minimize the packet loss, but can result in service degradation, since it may cause the substantial amount of packet loss when the period of scan operation is very short. As mentioned earlier, accessing one wireless channel during one scan operation may not good for the networking environments in which the APs are not configured densely. In order to solve these problems, we introduce the concept of scan granularity which enables the number of wireless channels to be accessed during a scan operation to be varied for increasing the integrity of neighboring AP information. The integrity of neighboring AP information indicates the degree of goodness for the neighboring AP information, i.e., whether it contains the detailed, complete, and correct information about neighboring APs or not, when the mobile nodes moves.

[71]

[72] Definition 3: The scan granularity is defined as the number of wireless channels which are accessed during a scan operation.

[73]

[74] Since the time to access a wireless channel in wireless LAN is between 20 ms and 60 ms, and the application interrupt allowance time (AIAT) for VoIP application is about 50 ms, the scan granularity of VoIP should be one or at most two. For stored video/ audio applications, however, since the application interrupt allowance time is about a few seconds, the scan granularity can be 11. In other words, the full scan operation is proper.

[75]

[76] Lemma 4: In wireless LAN, as the mobile node moves, the scan granularity for maximizing the integrity of neighboring AP information, while minimizing the power consumption due to the handover procedure is 11 (or 13).

[77] Proof: In order to minimize the power consumption due to the handover procedure, it is necessary to access as many as possible channels during a scan operation and getting as much as possible neighboring AP information. This is achieved by the full scan operation. Since the number of wireless channels in wireless LAN is 11 (or 130, the scan granularity for maximizing the integrity of neighboring AP information, while minimizing the power consumption due to the handover procedure is 11 (or 13). D

[78]

[79] Theorem 3: As a mobile node moves in a wireless LAN, the handover procedure with both minimal power consumption and maximal integrity of neighboring AP information while sustaining the service quality of an application is performing a full scan operation during T sec.

[80] Proof: It is clear from Theorem 1 and Lemma 4. D

[81]

[82] Up to now, we have done some theoretical analysis for performing optimal handover procedure with respect to power consumption, service quality, packet loss, and the integrity of neighboring AP information. However, in reality, Q(t), i.e., the input rate of the application buffer, depends on the network conditions. So, it is desirable to make the scan period to be adaptively changed in accordance with the network conditions, in order to minimize both the power consumption and packet loss while preserving the service quality during handover. In a dense networking environment such as indoors and large metropolitan area, the small scan granularity is desirable. This is because the sufficient AP information can usually be obtained even though a few channels are only accessed during a scan operation in dense networking environments.

[83] It is found that performing a full scan during T sec is an effective handover procedure for minimal power consumption and maximal integrity of neighboring AP information, if the packet loss can be recovered without time delay at application layer. However, the full scan operation may incur large packet loss during handover operation, so that the application buffer may not be filled in after handover. This may result in the severe service degradation. Therefore, it is necessary to dynamically adjust the scan granularity in accordance with the network conditions, in such way that both power consumption and packet loss are minimized while preserving the service quality during handover. In addition to that, as described before, accessing only one channel during a scan operation may result in poor neighboring AP information. Specially, if there is no AP existing for accessed channel, or the RSSI is very weak from the AP in accessed channel, the mobile node may not perform handover even though the signal strength from the current AP is very low. Therefore, there is a tradeoff between minimization of power consumption and packet loss, and maximization of the integrity of neighboring AP information.

[84] Fig. 4 shows the application-aware handover algorithm which can find the optimal next AP with both minimal power consumption and minimal packet loss, while preserving the service quality during handover, taking into consideration the networking conditions. The algorithm in Fig. 4 consists of three parts; identifying the type of application and finding the neighbor APs, the handover preparation, and channel switching for AP handover. The identification of the type of application is described previously in Fig. 3, in detail. In the identification of the type of application, the main application which is currently running on the mobile node is identified, and the values of application interrupt allowance time and scan granularity are determined. In particular, the end-to-end delay time and allowable packet loss are taken into account for determining the proper value of the scan granularity in order to preserve the service quality of the application during handover.

[85] In Theorems 1-3, it is found that the scanning method achieving the minimality in both power consumption and packet loss while preserving the service quality of the application is to scan one wireless LAN channel during the maximum application interrupt time T sec. The application-aware handover algorithm in Fig. 4 takes advantage of these properties of the scan operation, so that it tries to perform one scan operation, every T sec, as long as the application buffer contains enough pre-fetched data to preserve the service quality. However, the input data rate at an application buffer depends on the network conditions. It usually varies with time t, so that the amount of data which has been filled up at the application buffer during T sec may be less than Q. In this case, the algorithm prolongs the scan operation to the multiple of T sec until the application buffer fills up to Q. Therefore, the scanning period varies in accordance with the network conditions, with the value of N*T seconds where N is an integer and usually less than 5. N is also decided by the application type.

[86] The operation of application- aware handover algorithm in Fig. 4 is as follows. It consists of three phases; Searching of neighboring AP information, Handover preparation, and handover execution. These phases are determined by two signal strength threshold values Tp and Ts, where Tp is called the handover preparation threshold, and Ts is called the handover execution threshold. First, during initialization, the algorithm decides the values of T, Q and G where T, Q and G are the maximum of application interrupt allowance time, application data to play out, and scan granularity, respectively. These values are determined by the type of application.

[87] During the phase of searching the neighboring AP information, if the moving average of RSSI value from the current AP is greater than Tp, the mobile node first scans the G number of channels, getting the neighboring AP information, and stores them into Neighbor_AP_List file. Next, it waits for another T seconds, and start scanning operation again for another G number of channels, if the amount of pre- fetched data is larger than or equal to Q bytes. These steps are repeated until all the channels are scanned. After all the channels are scanned, it starts scanning operation again from the first set of G channels. However, for each scanning period, if the amount of data filled up at the application buffer is less than Q, the scanning period is prolonged until the buffer is filled up sufficiently enough to preserve the service quality of the application. In this way, during the searching phase, the service quality of an application is preserved with minimal power consumption due to handover operation.

[88] As the moving average value of RSSI becomes less than Tp due to the movement of the mobile node, the mobile node gets into the handover preparation phase. During the handover preparation phase, higher priority is put on finding optimal next AP and performing fast handover. So, the mobile node first selects candidates for next AP from the Neighbor_AP_List, and stores them into Candidate_AP_List. The APs with good AP/Channel load, and whose RSSIs are greater than Ts are selected for candidate APs. AP/Channel load conditions mean the channel load and traffic load which are defined in IEEE 802. I l k standard. The scan granularity G is adjusted to achieve fast handover with minimal packet loss. If the RSSI from the current AP is greater than Ts, the mobile node performs, with reduced scanning period, the scan operation for the APs in Candidate_AP_List, and updates the candidate AP information. In this way, both the scan period and scan granularity are getting smaller, so that the mobile node can switch to the optimal next AP with minimal handover latency.

[89] When the signal strength from the current AP becomes less than Ts, the mobile node gets into the handover execution phase. In the handover execution phase, the mobile node selects the optimal next AP from the Candidate_AP_List which has the strongest RSSI values and best AP/Channel load conditions, and rapidly switch to it for the completion of the handover procedure. The structures of the above the Neighbor_AP_List and Candidate_AP_List are shown in Fig. 5. The Neighbor_AP_List and Candidate_AP_List are usually stored in cache memory for fast access.

[90] The salient feature of the handover algorithm in Fig. 4 is that the service quality of the application is guaranteed during the movement of a mobile node by dynamically adjusting the scanning period in a manner that it is set to T when there is enough data flowing into the application buffer through network, and it is otherwise set to larger value than T. In other words, as a mobile node gathers the neighboring AP information during movement, the scan period is adaptively changed in accordance with the network condition, so that the degradation of service quality of the application can be prevented.

[91] Another salient feature of the handover algorithm in Fig. 4 is that during neighboring

AP information gathering step and handover preparation step, it uses, in harsh communication environment with fluctuating electrical waves, the estimate of moving average of signal strength instead of RSSI to estimate the mobility pattern. In this way, it can gather more accurate neighboring AP information by reducing estimation error for handover in harsh wireless communication environment such as indoor and/or metropolitan area.

[92] Another salient feature of the handover algorithm in Fig. 4 is that in addition to signal strength, the channel and AP load condition defined in IEEE 802.11 k are taken into account as criteria for the selection of candidate AP.

[93] Another salient feature of the handover algorithm in Fig. 4 is that when determining the scan period during the handover preparation step, if the neighboring APs are densely installed, the small value of scan granularity is chosen, and otherwise the large value of scan granularity is chosen. This can not only reduce the possible packet loss due to the scanning operation, but also achieve faster handover by selecting optimal AP.

[94] And, the scan period is reduced either by the amount of T/K sec per scan, or in a sequence <I1 sec, 12 sec, ..., In sec> where K and Ii are real values with Il>12>...>In. In this way, during the handover preparation period, the scan period and scan granularity are adaptively adjusted to the type of application and wireless networking environment in a manner that the degradation of application service, packet loss, and handover latency are minimized.

[95] Another salient feature of the handover algorithm in Fig. 4 is that when the RSSI from the current AP is less than or equal to handover execute threshold Ts, the mobile node enters into the handover execution step. During the handover execution step, the next AP is selected from the candidate APs which has the optimal performance in both signal strength and AP load condition, and the point of attachment is changed to the next AP, and the handover operation is finally completed.

[96] In summary, the salient features of the proposed application- aware and power- efficient handover algorithm is that the scan period and scan granularity are adaptively changed in accordance with the type of the application and network conditions in a manner the degradation of service quality and power consumption are minimized, when the mobile node moves between different APs.

[97] Now, we describe an example by using a figure. Fig. 6 shows an example of the application-aware and power-efficient handover algorithm at wireless LAN. In Fig. 6, as the mobile node moves, the estimate of moving average of signal strength from the current AP is larger than Tp, the pre-scanning is performed with the integral multiple of the maximum AIAT. In this case, the scan granularity is usually set to be that of the full scan. During AP information gathering step, the neighboring AP information is gathered without affecting the service quality of the application, so that it is said to be in the region of guaranteed service quality. As described before in Theorems 1 3, the power consumption due to the pre-scanning is minimized.

[98] As the mobile node moves away from the current AP, the signal strength from the current AP is also decreasing. If the estimate of moving average of signal strength from the current AP is less than or equal to Tp, the mobile node enters into the handover preparation step, in which the goal is to handover to the optimal AP, rather than maintaining the service quality of the application. So, it is said to be in the region of optimal AP selection. In order to achieve the goal, the mobile node select candidate APs from the neighboring APs, whose signal strengths are beyond some specific value, and whose AP/channel loads are not large. The scan period and scan granularity are adjusted to be smaller in order to find the optimal AP. However, if the scan period is abruptly set to be small value, the service disruption may occur, so that the scan period is reduced either in a step-wise manner, or a specific sequence. In Fig. 6, the scan period is reduced from 2T to T, and T to T/2, and the scan granularity is also reduced. In this way, the variation of signal strength from the candidate APs could be very precisely found out, which may in turn result in fast handover to the optimal AP.

[99] [100]

Industrial Applicability

[101] As described above, the present invention can provide stable, power-efficient and fast handover to support QoS-guaranteed real-time mobile multimedia communication service such as mobile VoIP at indoor or metropolitan area with harsh electric wave characteristics.

[102] According to the recent on the wireless LAN usage pattern, more than 70 % of

WLAN mobility occurs within campus, office or factory area in which APs are usually densely installed, the applicability of the present invention is very large. This is because the handover method and system of the present invention is particularly useful in complex wireless networking environment such as indoors and/or large metropolitan area in which APs are densely installed.

[103] The present invention is focusing on the efficient handover for the WLAN based on IEEE 802.11 a/b/g/n international standards. However, it can also be applied to other similar wireless networks including IEEE 802.16 Wireless MAN.

[104]

[105]

[106]

Claims

[1] An application- aware and power-efficient fast handover method in a wireless

LAN, comprising the steps of: in order to guarantee the service quality of an application and perform power-efficient handover, by enabling handover procedure to consist of three steps of neighboring AP information gathering, handover preparation, and handover execution, using two handover threshold values Tp and Ts; and when the mobile node moves between different APs, during neighboring AP information gathering step, initialize the maximum value of application interrupt allowance time T, scan granularity G and the service initiation application data Q, and when the signal strength from the current AP is larger than the handover prepare threshold Tp, gather periodically neighboring AP information with variable scan period without degrading the service quality of the application while the power consumption for pre-scan being minimized, and registers the gathered AP information in neighboring AP information file; and when the mobile node moves between different APs therein, entering into the handover preparation step when the signal strength from the current AP is less than or equal to the handover prepare threshold Tp, select the candidate APs from the APs in neighboring AP information file, which have the outstanding performance in both signal strength and AP/load, and store them in candidate AP information file, and adjusting again the scan period and scan granularity in accordance with the type of the application and network condition, by doing scan operation again for the candidate AP, find out the optimal AP from the candidate APs which has the best performance; and when the mobile node moves between different APs therein, entering into the handover execution step when the signal strength from the current AP is less than or equal to the handover execute threshold Ts, change the point of attachment from the current AP to the optimal AP, and complete the handover procedure.

[2] The application-aware and power-efficient fast handover method of claim 1, wherein in the step of gathering periodically neighboring AP information with variable scan period without degrading the service quality of the application while the power consumption for pre-scan being minimized, and registering the gathered AP information in neighboring AP information file, if the signal strength from the current AP is larger than Tp, scan G channels from the entire channels, and store the scanned AP information into neighboring_AP_info_file, and this procedure is repeatedly performed with period T until the application buffer is filled up with Q amount of data, and after the entire channel is scanned, the scanning procedure starts again by scanning G channels, and if the amount of data filled up in application buffer during T is less than Q due to the network condition change, the scan period is elongated to the multiple of T until the buffer is filled up with Q to avoid the service degradation.

[3] The application-aware and power-efficient fast handover method of claim 1, wherein in the step of adjusting again the scan period and scan granularity in accordance with the type of the application and network condition, by doing scan operation again for the candidate AP, and finding out the optimal AP from the candidate APs which has the best performance, select, among the APs which have been stored in neighboring_AP_info_file, those whose signal strengths are larger than Ts and AP/channel load performance are outstanding, and store them into candidate_AP_info_file, and measure the signal strength from the current AP, and if it is larger than Ts, reduce the scan period and scan granularity by taking into account the type of application and network condition, and perform scanning operation to APs in candidate_AP_info_file, and thereby reducing the handover latency, and simultaneously finding out the optimal AP.

[4] The application-aware and power-efficient fast handover method of claim 3, wherein in the step of reducing the scan period and scan granularity by taking into account the type of application and network condition, in the process of determining the scan period, if the neighboring APs have been densely installed, reduce the scan granularity, and otherwise increase the scan granularity, which allows the optimal AP to be selected while reducing packet loss, thereby achieving fast handover, and by making the scan period to be reduced either in specific sequence <I1 sec, 12 sec, ..., In sec> with real number Ii such that Il>12>...>In, or T/K with increasing K per scan operation where K is an integer, and the probability of finding out the optimal AP in a given time is increased, which prevents the degradation of service quality of the application, and minimizes the handover delay.

[5] The application-aware and power-efficient fast handover method of claim 3, wherein in the step of selecting, among the APs which are stored in neighboring_AP_info_file, those whose signal strengths are larger than Ts and AP/channel load performance are outstanding, and storing them into candidate_AP_info_file, the neighboring_AP_info_file and the candidate_AP_info_file contains AP_id, MAC_addr, signal strength, AP/channel load, and authentication and security information.

[6] The application-aware and power-efficient fast handover method of claim 3, wherein in the step of gathering periodically neighboring AP information with variable scan period without degrading the service quality of the application while the power consumption for pre-scan being minimized, allow to use the estimate of moving average of the signal strength in the network environment with large fluctuating electric wave characteristics, and identify the mobility pattern, and get the more accurate neighboring AP information by reducing estimation error in indoor and/or large metropolitan area where the electric wave characteristics tend to abruptly change.

[7] An application- aware and power-efficient fast handover system in a wireless

[8] The application-aware and power-efficient fast handover system of claim 7, wherein in the step of gathering periodically neighboring AP information with variable scan period without degrading the service quality of the application while the power consumption for pre-scan being minimized, and registering the gathered AP information in neighboring AP information file, if the signal strength from the current AP is larger than Tp, scan G channels from the entire channels, and store the scanned AP information into neighboring_AP_info_file, and this procedure is repeatedly performed with period T until the application buffer is filled up with Q amount of data, and after the entire channel is scanned, the scanning procedure starts again by scanning G channels, and if the amount of data filled up in application buffer during T is less than Q due to the network condition change, the scan period is elongated to the multiple of T until the buffer is filled up with Q to avoid the service degradation.

[9] The application-aware and power-efficient fast handover system of claim 7, wherein in the step of adjusting again the scan period and scan granularity in accordance with the type of the application and network condition, by doing scan operation again for the candidate AP, and finding out the optimal AP from the candidate APs which has the best performance, select, among the APs which have been stored in neighboring_AP_info_file, those whose signal strengths are larger than Ts and AP/channel load performance are outstanding, and store them into candidate_AP_info_file, and measure the signal strength from the current AP, and if it is larger than Ts, reduce the scan period and scan granularity by taking into account the type of application and network condition, and perform scanning operation to APs in candidate_AP_info_file, and thereby reducing the handover latency, and simultaneously finding out the optimal AP.

[10] The application-aware and power-efficient fast handover system of claim 9, wherein in the step of reducing the scan period and scan granularity by taking into account the type of application and network condition, in the process of determining the scan period, if the neighboring APs have been densely installed, reduce the scan granularity, and otherwise increase the scan granularity, which allows the optimal AP to be selected while reducing packet loss, thereby achieving fast handover, and by making the scan period to be reduced either in specific sequence <I1 sec, 12 sec, ..., In sec> with real number Ii such that Il>12>...>In, or T/K with increasing K per scan operation where K is an integer, and the probability of finding out the optimal AP in a given time is increased, which prevents the degradation of service quality of the application, and minimizes the handover delay.

[11] The application-aware and power-efficient fast handover method of claim 9, wherein in the step of selecting, among the APs which are stored in neighboring_AP_info_file, those whose signal strengths are larger than Ts and AP/channel load performance are outstanding, and storing them into candidate_AP_info_file, the neighboring_AP_info_file and the candidate_AP_info_file contains AP_id, MAC_addr, signal strength, AP/channel load, and authentication and security information.

[12] The application-aware and power-efficient fast handover method of claim 9, wherein in the step of gathering periodically neighboring AP information with variable scan period without degrading the service quality of the application while the power consumption for pre-scan being minimized, allow to use the estimate of moving average of the signal strength in the network environment with large fluctuating electric wave characteristics, and identify the mobility pattern, and get the more accurate neighboring AP information by reducing estimation error in indoor and/or large metropolitan area where the electric wave characteristics tend to abruptly change.