TWI419515B - The network bandwidth upper limit statistics and adjustment method of network equipment - Google Patents

Description

Network bandwidth upper limit statistics and adjustment methods for network devices

The invention relates to a network bandwidth upper limit statistics and adjustment method for a network device, which can correct and adjust the mis-estimated bandwidth upper limit value without modifying the previous bandwidth detection technology.

Today's network devices, such as broadband routers or wireless network broadband sharers, use network bandwidth management (QoS) and service types (Type) for the network traffic, bandwidth allocation and restriction functions of the device itself. Of Service (TOS) mechanism, and uses priority technology to set bandwidth allocation and restriction for traffic of various protocols or traffic of a specific endpoint. In addition, the bandwidth allocation and restriction functions of these network devices must know the upper limit of the external bandwidth in advance to allocate a bandwidth for each traffic. However, many of today's bandwidth detection techniques, due to differences in their algorithms or network deployment conditions, can cause errors in the detection of actual bandwidth, such as overestimation or underestimation.

In the past, network devices often used a Round Trip Time (RTT) to estimate the network bandwidth, and transmitted packets for specific endpoints (for example, DNS, GATEWAY, or NTP server), and received the time difference of the response packets. Mathematical operations to estimate the bandwidth. However, the above techniques and practices may cause large variations in error range due to network conditions and network lengths in specific endpoints. For example, when the network is congested, a lower bandwidth may be estimated ( 2Mbps bandwidth only estimates 1Mbps bandwidth), the QoS mechanism uses the lower bandwidth value (1Mbps) for bandwidth allocation and limitation, which will cause the user to use only 1Mbps bandwidth, resulting in Waste of bandwidth usage. On the other hand, if a higher bandwidth value is estimated (2 Mbps bandwidth is estimated to be 3 Mbps bandwidth), when QoS performs bandwidth allocation and limitation, it is assumed that a high priority allocation is given for 2 Mbps, and a low priority is given for allocation of 1 Mbps. A priority of 1 Mbps in the traffic bandwidth of 2 Mbps will be occupied by low priority, which will make the high priority only 1 Mbps traffic bandwidth at full load. In this way, the actual bandwidth will not be allocated according to the original set ratio.

In the face of the above-mentioned error caused by bandwidth detection, the previous method can only modify the calculation method of the bandwidth detection algorithm or design different detection algorithms to improve the error of the detection as much as possible, but Accurate algorithms or detection methods still have unacceptable error ranges or are inaccurate detection conditions and environmental conditions, and the results of bandwidth detection lack accuracy.

The main purpose of the present invention is to provide a network bandwidth upper limit statistics and adjustment method for a network device, which does not need to modify the previously proposed bandwidth detection technology to detect the network detected by the previous bandwidth detection technology. The path bandwidth value is introduced into the statistics and adjustment method of the present invention, and the upper limit value of the estimated error bandwidth can be corrected and adjusted.

A secondary object of the present invention is to provide a network bandwidth upper limit statistics and adjustment method for a network device, which can be applied to an intranet, an external network, a wired network, or a wireless network. The wide upper limit is adjusted to meet the needs of various network systems.

To this end, the present invention provides a network bandwidth upper limit statistics and adjustment method for a network device, the steps comprising: obtaining a current upper bandwidth upper limit value in the network device; setting a sampling amount and an acceptance ratio value to Deciding to accept the threshold value; providing a statistical table for recording the number of statistics of the plurality of index values, each index value corresponding to a bandwidth class, and each of the bandwidth levels is different from the first-order unit; performing a bandwidth upper limit The statistical process includes the following steps: obtaining a network bandwidth value detected by the bandwidth detection technology currently running in the network device and recording a sampling number; converting the network bandwidth value into an index value, according to the network The index value converted by the road bandwidth value is incremented by 1 for the index value of the equivalent value in the statistical table, and 1 is added to the sampling number; and the comparison sampling number reaches the sampling amount, and if the sampling frequency is not reached The sampling amount is continued, and the bandwidth upper limit statistical process is continued. Otherwise, a bandwidth upper limit adjustment process is performed; and the bandwidth upper limit adjustment process is performed, and the steps include: the upper limit of the conversion bandwidth is an index value. The number of statistics of the index value of the same value is taken from the statistical table according to the index value converted from the upper limit value of the bandwidth; and whether the number of statistics taken is greater than the acceptance threshold, and if the number of statistics taken is greater than the acceptance threshold , the next time the upper limit of the bandwidth is counted again. Otherwise, the index value of the most statistics is determined in the statistics table, and then the index value converted by the maximum number of statistics and the index value converted by the upper limit of the bandwidth are compared. The value of the value is decremented by the index value converted from the upper limit of the bandwidth, plus 1 or up to the present value and multiplied by the pitch unit to calculate the updated bandwidth upper limit instead of the current frequency. Wide upper limit and After updating the upper limit of the bandwidth, the next upper bandwidth upper limit statistics flow is performed again.

Please refer to FIG. 1 and FIG. 2 , which are flowcharts of a preferred embodiment of a network bandwidth upper limit statistics and adjustment method for a network device according to the present invention, and a statistical table of an embodiment of the present invention. . The network bandwidth upper limit statistics and adjustment method of the network device of the present invention can be applied to a network device, and can be counted by using the network bandwidth detected by the previously proposed bandwidth detection technology. In order to adjust the upper limit of the more accurate bandwidth, the bandwidth is smoothly distributed.

As shown in FIG. 1 , the network bandwidth upper limit statistics and adjustment method of the present invention includes the following main steps: First, in step S111, a network bandwidth statistical adjustment is performed in the network device when the network device starts to operate. The resident program will start and obtain the current Upper Limit Bandwidth (ULBW) in the network device. The upper bandwidth limit (ULBW) can be manually determined by the user or through a previously proposed The bandwidth detection technology is automatically developed.

In step S112, a total number S of samples and an acceptable ratio value R (a floating point number between 0 and 1) are set to determine an acceptance threshold A=R×S.

Next, in step S113, a statistical table is provided. The statistical table is used to record the statistical number T[I _N ] of the plurality of index values I _N , and each index value I _N corresponds to a bandwidth level, and the bandwidth is different from the first level. Unit, this embodiment uses 128Kbps as the span level unit of the bandwidth hierarchy, the index value I ₀ corresponds to 0Kbps, the index value I ₁ corresponds to 128Kbps, the index value I ₂ corresponds to 256Kbps, and the index value I ₃ corresponds to 384Kbps... (Legacy is like this). Of course, in another embodiment of the present invention, different unit values may be used as the level unit of the bandwidth level, for example, 32 Kbps, 64 Kbps, 256 Kbps, 512 Kbps, or any unit value. Thereafter, a bandwidth upper limit statistics flow 120 is entered.

The bandwidth upper limit statistics flow 120 of the present invention is as follows: Step S121, sampling a network bandwidth (NBW), and obtaining the network bandwidth value (NBW) Use general bandwidth detection techniques currently running on network devices, such as packet round-trip time RTT technology (packet size/time difference), traffic statistics techniques (collect packet flow/collection time), or other previously proposed frequencies The wide detection technique simultaneously records a variable C of the number of samples.

Step S122, the network bandwidth value (NBW) is converted into a corresponding index value I _A according to a first bandwidth step equation, and the equation (1) is as follows: I _A = (NBW+ Mod A (1) where I _A is the index value obtained by converting the network bandwidth value (NBW), NBW is the network bandwidth value, mod is the division quotient operation function, and A is the step unit ( Such as 128Kbps). Value by network bandwidth (NBW) as an index value in terms of I _A to the value equivalent to the statistical tables of index values which count the number of T I _N I _N [1] add up, for example: detect network When the bandwidth value (NBW) is 963 Kbps, the index value I _A = (963 + 64) mod 128 = 8 can be calculated by the equation (1), and the statistical number T of the statistical table I _{N = 8} in the second graph [ I ₈ 〕 plus 1 is added, and at the same time, the number C of sampling times for the network bandwidth value (NBW) is also increased by 1 (C=C+1).

Next, in step S123, if the sampling frequency C of the current network bandwidth value (NBW) does not reach the sampling amount S, the bandwidth upper limit statistical process 120 is continued until the sampling frequency C is the same as the sampling amount S, and then the frequency is performed. The upper limit adjustment process 130.

After the bandwidth upper limit statistics process 120 collects a sufficient number of samples for the network bandwidth value (NBW), the bandwidth upper limit adjustment process 130 of the present invention is entered, and the flow is as follows: step S131, bandwidth The limit value (ULBW) is converted into a corresponding index value I _B according to a second bandwidth step equation, and the equation (2) is as follows: I _B = (ULBW) mod A (2) where I _B is a conversion The index value obtained by the upper bandwidth limit value (ULBW), ULBW is the upper limit of the bandwidth, mod is the division function of the division, and A is the unit of the step (for example, 128 Kbps). The index value I _B converted by the upper limit value of the bandwidth (ULBW) is used to extract the index value I _{N of} the same value from the statistical table, and the statistical number T[I _N ] of the record, for example, when the bandwidth upper limit value ( When ULBW is defined as 896 Kbps, the converted index value I _B=7 , the statistical number of I _N=7 is taken from the statistical table of Fig. 2 [I ₇ ]=7; when the upper limit of the bandwidth (ULBW) When the index value I _{B=8 is} calculated as 1024Kbps, the statistical number of I _N=8 is taken from the statistical table of Fig. 2[I ₈ ]=81; when the upper limit of the bandwidth (ULBW) When the index value I _{B=9 is} calculated as 1152 Kbps, the statistical number T[I ₉ ]=4 of I _N=9 is taken from the statistical table of FIG. 2 .

In step 132, after the statistical number T[I _N ] is taken out, the comparison statistical number T[I _N ] is greater than the acceptance threshold A. If the number of statistics T[I _N ] is greater than the acceptance threshold A, the next bandwidth upper limit statistics flow 120 is performed again to perform the bandwidth statistics of the next cycle. On the other hand, if the comparison statistics T[I _N ] is less than the acceptance threshold A, the upper bandwidth limit (ULBW) may make an error or an estimation error, and the index value I _{N of the} upper bandwidth limit (ULBW) is counted. The super-state distribution in the table, followed by the relevant steps of adjusting the current upper bandwidth limit (ULBW), will again perform the bandwidth upper limit statistics flow 120.

As an example, if the total number of samples S is set to 100 times of sampling and the accepted ratio R is 0.8, the receiving threshold A can be determined as 80 times, when the upper limit of the bandwidth (ULBW) is converted. For the index value I _B=8 , and the statistical number T[I ₈ ]=81 corresponding to the extracted index value I _N=8 in the statistical table of FIG. 2 will exceed the acceptance threshold A, which is currently estimated or formulated. The upper bandwidth limit (ULBW) is a more accurate upper limit value, and can be returned to the upper bandwidth upper limit statistical process 120 without adjustment; conversely, when the upper limit of the bandwidth (ULBW) is converted into an index value If I _{B is} not equal to 8 and is other values, then the statistical number T[I _N ] of the index value I _N taken out in the statistical table of FIG. 2 must be less than the acceptance threshold A, which is currently estimated or formulated. The upper bandwidth limit (ULBW) may also be the upper limit of the overestimation, underestimation or misestimation. The upper bandwidth limit (ULBW) must be further adjusted, and after the adjustment is completed, the bandwidth upper limit statistics flow 120 is returned. .

In step 133, when the statistical number T[I _N ] of the extracted index value I _N is less than the acceptance threshold A, the index value I _N of the maximum number of statistics is determined from the statistical table of FIG. 2 ( The index value of the maximum number of sampling times in this embodiment is I _N=8 , and the number of statistics is T[I ₈ ]=81).

In step 134, the numerical value between the index value I _N of the maximum number of statistics T[I _N ] and the index value I _B converted by the bandwidth upper limit value (ULBW) is compared. If the index value I _{N of the} maximum number of statistics T[I _N ] is smaller than the index value I _B , then I _{B is} decremented by 1 (I _B =I _B -1) and multiplied by the rank unit (eg 128 Kbps) to replace the current The upper bandwidth limit (ULBW = (I _B -1) × 128 Kbps), in other words, the current upper bandwidth limit (ULBW) is overestimated, and the first-order bandwidth level must be lowered. If the index value I _{N of the} maximum number of statistics T[I _N ] is greater than the index value I _B , then I _{B is} incremented by 1 (I _B =I _B +1) and multiplied by the rank unit (eg 128 Kbps) to replace the current The upper bandwidth limit (ULBW = (I _B +1) × 128 Kbps), in other words, the current upper bandwidth limit (ULBW) is underestimated, and the first-order bandwidth level must be raised. Of course, in step 134, the instability of the bandwidth quality may also cause a misclassification of the upper bandwidth limit (ULBW), and the index value I _{N of the} most statistical number T[I _N ] is equal to the index value I. _B , however, the maximum number of statistics T[I _N ] is still less than the set acceptance threshold A. At this time, it is not necessary to adjust the upper bandwidth limit (ULBW) to avoid the network with unstable bandwidth. The device adjusts the upper bandwidth limit (ULBW).

After the step 134 is performed, the next period bandwidth upper limit statistics process 120 and the bandwidth upper limit adjustment process 130 are performed again, and the bandwidth value detected by the network device is continuously counted, and periodically for the network device. The upper bandwidth limit (ULBW) is adjusted.

In addition, the total number of samples S in the embodiment of the present invention is taken as an example of one hundred sampling times. However, in order to increase the concentration of the sampled network bandwidth (NBW), it may be further set to one hundred times. , one thousand times, ten thousand times or other higher number of sampling times, and the proportional value R increases with the accuracy of the sampling requirement, the higher the proportional value R is set and is close to 1, the precision of sampling is also required. Relatively improved.

Furthermore, the bandwidth upper limit statistics and adjustment method of the present invention can be used for various wired network interfaces (eg, Ethernet interfaces) or wireless network interfaces included in network devices of external and internal network lines ( For example: 802.11a, 802.11b/g) network bandwidth value (NBW) for detection and statistics, and bandwidth upper limit (ULBW) can be a bandwidth The upper limit of the upload (Uplink Upper Limit Bandwidth; UULBW) or the upper limit of the bandwidth download (Downlink Upper Limit Bandwidth; DULBW).

In summary, the present invention uses the statistical principle to relatively record the network bandwidth estimated by each of the previous various bandwidth detection techniques or algorithms, and adjusts and updates the network device through cyclic analysis and comparison every statistical period. The bandwidth value of the running medium improves the error of various bandwidth detection technologies to obtain the upper limit (ULBW) of the network device physical line bandwidth which is more accurate in the normal state distribution.

The above is only the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention, that is, the variations, modifications, and modifications of the shapes, structures, features, and spirits described in the claims of the present invention. All should be included in the scope of the patent application of the present invention.

120. . . Bandwidth upper limit statistical process

130. . . Bandwidth upper limit adjustment process

FIG. 1 is a flow chart of a preferred embodiment of a method for estimating and adjusting the bandwidth of a network device according to the present invention.

Figure 2 is a statistical table of an embodiment of the statistical number of index values of the present invention.

120. . . Bandwidth upper limit statistical process

130. . . Bandwidth upper limit adjustment process

Claims (10)

A network bandwidth upper limit statistics and adjustment method for a network device, the steps comprising: obtaining a current upper bandwidth upper limit value in the network device; setting a sampling amount and an acceptance ratio value to determine an acceptance threshold value Providing a statistical table for recording statistics of a plurality of index values, each index value corresponding to a bandwidth class, and each of the bandwidth levels is a first-order distance unit; performing a bandwidth upper limit statistical process, steps The method includes: sampling a network bandwidth value detected by a bandwidth detection technology currently running in the network device, and recording a sampling number; converting the network bandwidth value to the index value, according to the network The index value converted by the road bandwidth value is incremented by 1 for the index value of the same value in the statistical table, and 1 is added to the sampling number; and the sampling time is compared to whether the sampling amount reaches the sampling amount. If the sampling number does not reach the sampling amount, continue the bandwidth upper limit statistical process; otherwise, perform a bandwidth upper limit adjustment process; and perform the bandwidth upper limit adjustment process, the steps include: converting the frequency The upper limit value is the index value, and the index value of the same value is taken from the statistical table according to the index value converted from the upper limit value of the bandwidth, and the statistical number of the record is recorded; and whether the counted number of statistics is compared Greater than the acceptance threshold, if the counted number of statistics is greater than the acceptance threshold, re-run The bandwidth upper limit statistics flow is next, otherwise, the index value of the maximum number of statistics is determined in the statistics table, and then converted by the index value of the maximum number of statistics and the upper limit value of the bandwidth. The value between the index values is decremented by 1 for the index value converted to the upper limit of the bandwidth, plus 1 is added or the present value is maintained and multiplied by the rank unit to calculate the updated The bandwidth upper limit value is substituted for the current upper bandwidth limit value, and after updating the bandwidth upper limit value, the next upper bandwidth upper limit statistical flow is performed again. The network bandwidth upper limit statistics and adjustment method described in claim 1, wherein when the index value of the maximum number of statistics is less than the index value converted by the upper bandwidth limit, the bandwidth is The index value converted by the limit value is decremented by one, and when the index value of the maximum number of statistics is greater than the index value converted by the upper limit value of the bandwidth, the upper limit value of the bandwidth is converted. The index value is incremented by one, and when the index value of the maximum number of statistics is equal to the index value converted by the upper limit of the bandwidth, the index value converted by the upper limit of the bandwidth maintains the present value. . For example, the network bandwidth upper limit statistics and adjustment method described in claim 1 wherein the pitch level unit is 32 Kbps, 64 Kbps, 128 Kbps, 256 Kbps, 512 Kbps or any unit value. For example, the network bandwidth upper limit statistics and adjustment method described in claim 3, wherein the network bandwidth value is converted according to a first bandwidth step equation, the first bandwidth step equation As follows: Where I _{A is} the index value, NBW is the network bandwidth value, mod is the division quotient operation function, and A is the step size unit. The network bandwidth upper limit statistics and adjustment method described in claim 3, wherein the bandwidth upper limit value is converted according to a second bandwidth step equation, the second bandwidth step formula As shown below: I _B =ULBW mod A where I _{B is} the index value, ULBW is the upper limit of the bandwidth, mod is the division quotient operation function, and A is the step unit. The network bandwidth upper limit statistics and adjustment method described in claim 1, wherein the upper limit of the bandwidth is an upper limit of a bandwidth upload or an upper limit of a bandwidth download. For example, the network bandwidth upper limit statistics and adjustment method described in claim 1 wherein the sampling amount is set to one hundred, one thousand, ten thousand or other number of sampling times. For example, the network bandwidth upper limit statistics and adjustment method described in claim 1 wherein the bandwidth upper limit statistics and adjustment method periodically counts the network bandwidth value of the network device and the bandwidth. The limits are adjusted. For example, the network bandwidth upper limit statistics and adjustment method described in claim 1 for the upper/download bandwidth upper limit of one of the network devices or the upper/download bandwidth of a pair of intranets The upper limit is adjusted. The network bandwidth upper limit statistics and adjustment method according to claim 1, wherein the network device comprises at least one wired network interface and/or at least one wireless network interface, and the network bandwidth upper limit statistics And the adjustment method is applied to the wired network and/or the wireless network connected to the network device.