CN114710807A - 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method - Google Patents

5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method Download PDF

Info

Publication number
CN114710807A
CN114710807A CN202210463305.0A CN202210463305A CN114710807A CN 114710807 A CN114710807 A CN 114710807A CN 202210463305 A CN202210463305 A CN 202210463305A CN 114710807 A CN114710807 A CN 114710807A
Authority
CN
China
Prior art keywords
rate
congestion
kalman filtering
data packet
available bandwidth
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210463305.0A
Other languages
Chinese (zh)
Inventor
王敏
罗义
袁凌云
王俊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Yunnan Normal University
Original Assignee
Yunnan Normal University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yunnan Normal University filed Critical Yunnan Normal University
Priority to CN202210463305.0A priority Critical patent/CN114710807A/en
Publication of CN114710807A publication Critical patent/CN114710807A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/0289—Congestion control
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/50—Reducing energy consumption in communication networks in wire-line communication networks, e.g. low power modes or reduced link rate

Abstract

The invention relates to a Kalman filtering dynamic congestion control method for 5G millimeter wave communication, and belongs to the technical field of network transmission communication. Aiming at the characteristics of high bandwidth, low time delay and high dynamic property in 5G millimeter wave communication, the congestion window adjustment strategy based on available bandwidth is adopted, the Kalman filtering algorithm is utilized to estimate the actual available bandwidth of a link, and then the stepping rate (pacingrate) and the congestion window (cwnd) are set according to the estimation value, so that the bandwidth utilization rate is effectively improved. Meanwhile, the rapid congestion sensing algorithm is used for detecting network congestion in time, the increase rate of a congestion window is controlled as early as possible before a network congestion point appears, congestion is effectively relieved, the queue length is reduced, round-trip delay is obviously reduced, and efficient transmission of data in a 5G network is achieved. Simulation experiments prove that the invention realizes higher average throughput and lower round-trip delay in a 5G network scene.

Description

5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method
Technical Field
The invention relates to a Kalman filtering dynamic congestion control method for 5G millimeter wave communication, and belongs to the technical field of network transmission communication.
Background
As a new Generation broadband Mobile Communication Technology, a fifth Generation Mobile Communication Technology (5G) has the characteristics of high bandwidth, low latency, large connection, and the like, and is a network infrastructure for implementing man-machine interconnection. Millimeter wave (30-300GHz electromagnetic wave) communication brings speed and delay advantages to 5G, and simultaneously brings challenges to the design of transmission protocols at the upper layers of a 5G network, especially the design of a congestion control mechanism. Firstly, a TCP protocol widely used in the current network adopts a congestion control mechanism with slow start, additive increase and multiplicative reduction, so that the utilization rate of bandwidth is greatly limited, and bandwidth waste is caused in a 5G network with high bandwidth, therefore, how to improve the bandwidth utilization rate in the 5G network is a challenge faced by the congestion control mechanism; the existing congestion control mechanism takes packet loss as a signal of network congestion, packets in a network are temporarily stored in a route or a buffer of a base station before being discarded to form a queue, a large number of packets are queued in the buffer, and end-to-end transmission delay is increased, so that how to reduce the end-to-end transmission delay to highlight the advantage of low delay of a 5G network is the second challenge faced by the congestion control mechanism; the influence of external environment on 5G millimeter waves is very easy, the channel quality is very unstable, and thus high dynamic changes of bandwidth and time delay are caused, and how to quickly detect the changes of bandwidth and time delay to adapt to the high dynamics of the 5G network is the third challenge facing congestion control.
At present, a congestion control method for a 5G network is not enough in reporting ends, and through early experimental research, it is found that a BBR protocol proposed by google has good adaptability to the 5G network in an existing typical congestion control algorithm, but the BBR protocol has a slow response to network congestion and forms a long queue in a cache, so that the round-trip delay is significantly increased. In addition, the BBR cannot detect the change of the bottleneck bandwidth in time in a high-dynamic scenario, resulting in degraded throughput and round-trip delay performance.
Disclosure of Invention
The invention aims to solve the technical problem of providing a Kalman filtering dynamic congestion control method for 5G millimeter wave communication, aiming at the characteristic of high dynamics presented by the current 5G network, a Kalman filtering algorithm is used for accurately estimating the actual available bandwidth of a link, so that the sending rate of a sending end is better adapted to the dynamic change of the 5G link, and meanwhile, the response to network congestion is accelerated through a rapid congestion sensing algorithm, so that the purposes of improving the network throughput and reducing the round-trip delay are achieved.
The technical scheme of the invention is as follows: a Kalman filtering dynamic congestion control method for 5G millimeter wave communication is operated at a sending end of an end-to-end link. At the beginning of TCP connection establishment, a sending end enters a starting Stage (STARTUP), a congestion window cwnd and a stepping rate are exponentially increased, and a data packet is quickly sent to fill a link; meanwhile, a fast congestion sensing algorithm (FCA) is called to timely detect whether a link is fully utilized or not; when the link is detected to be fully utilized in the starting stage, entering a detection available bandwidth stage (PROBE _ AB), calling an end-to-end available bandwidth estimation algorithm based on Kalman filtering to estimate the actual available bandwidth of the current link, and setting cwnd and pacingrate according to the available bandwidth; and finally, the sending end controls the sending time of the next data packet according to the pacingrate, and the congestion caused by the burst data packet to the network is avoided.
The method comprises the following specific steps:
step 1: the sending end and the receiving end establish connection through TCP three-way handshake, and the sending end records rtt value during handshake.
The initialization variables of the transmitting end are as follows: congestion flag ispipeefilled ═ false, initial stepping rate ═ init _ cwnd ═ 8/rtt, previous rate-delay trace point prePoint ═ 0, 0 in fast congestion sensing algorithm, initial state vector x in Kalman filtering0＝[0.505，-0.5]TProcess noise covariance matrix in Kalman filtering
Figure BDA0003621164560000021
Estimation error covariance matrix in Kalman filtering
Figure BDA0003621164560000022
Figure BDA0003621164560000023
Noise covariance matrix R ═ 1 in Kalman filtering]。
Step 2: when the sending end sends a data packet p each time, recording the time packet, delayed _ time sent by the data packet and the total data amount packet, delayed, confirmed so far at the head of the data packet p, and calculating and storing the sending rate send _ rate of the data packet p according to a formula (1):
Figure BDA0003621164560000024
where p.size denotes the size of packet p,
Figure BDA0003621164560000025
indicating the transmission interval of packet p.
Step 3: when the sending end receives an ACK packet, the average transmission rate delivery _ rate of the data packet p confirmed by the ACK is calculated and stored according to the formula (2), the difference between the current system time and the sending time packet.
Figure BDA0003621164560000026
In the formula, delayed represents the size of a data packet successfully received currently, and the size of a data packet acknowledged by ACK is added every time a valid ACK is received, packet.
Step 4: the sending end selects the largest delivery _ rate value as the maximum transmission rate sample maxRate and selects the smallest rtt value as the smallest round-trip delay sample minRtt from the 10 sets of delivery _ rate and rtt values obtained recently.
Step 5: and the sending end judges whether the end-to-end link is fully utilized or not according to the congestion sign ispipeefilled, if the sign is true, the Step7 is entered, and if the sign is false, the Step6 is entered.
Step 6: the sender enters a start phase (STARTUP) and invokes a fast congestion sensing algorithm.
The Step6 comprises the following specific steps:
step6.1: entering a fast congestion sensing algorithm, and setting a current rate-delay track point (delivery _ rate, rtt) according to the average transmission rate delivery _ rate and round-trip delay rtt of the data packet p obtained in Step 3.
Step6.2: and judging whether the previous track point is empty, namely whether the delivery _ rate or rtt in the track point is 0, if at least one item is 0, entering Step6.5, and if not, entering Step6.3.
Step6.3: calculating the slope k of a straight line formed by the previous track point and the origin point, namely pre-pointk＝k*curPoint.delivery_rate。
Step6.4: judging whether the round-trip time delay curPoint of the current track point is greater than or equal to rttk. If yes, setting the congestion flag ispipeefilled to true, indicating that the available bandwidth of the link is fully utilized, and entering step 6.5. If not, directly entering Step6.5.
Step6.5: and finishing the quick congestion sensing algorithm, and updating the parameter pretoint ═ curPoint.
Step6.6: estimating maximum bandwidth delay product BDPmaxmaxRate minRtt. Setting congestion window cwnd ═ BDPmaxcwndGain, step rate pacingrte ═ maxRate ═ pacnggain.
Step 7: entering a stable probing available bandwidth stage (PROBE _ AB), and estimating the current available bandwidth by using an end-to-end available bandwidth estimation algorithm based on Kalman filtering.
The Step7 comprises the following specific steps:
step7.1: from the send _ rate in Step72 and the receive _ rate in Step3, the Kalman filtering measurement state z is obtainedkSend _ rate/receive _ rate-1 and transition matrix H ═ send _ rate, 1 in the measurement state and system state]。
Step7.2: the estimation process of the Kalman filtering algorithm comprises the following steps: using the initial state vector x in Step10And equation (3) obtaining the system state xEvaluation value
Figure BDA0003621164560000031
Obtaining a prior error covariance matrix of the prior estimate using equation (4)
Figure BDA0003621164560000041
Figure BDA0003621164560000042
Figure BDA0003621164560000043
Step7.3: the correction process of the Kalman filtering algorithm comprises the following steps: kalman gain K is obtained using equation (5)kThen using equation (6) to estimate a priori from the system state
Figure BDA0003621164560000044
And the actual measured value zkObtaining the posterior estimated value
Figure BDA0003621164560000045
Finally, the error covariance matrix P of the posterior estimated value is updated by using the formula (7)kTo update in the next iteration
Figure BDA0003621164560000046
In the formula (7), I is an identity matrix.
Figure BDA0003621164560000047
Figure BDA0003621164560000048
Figure BDA0003621164560000049
Step 7.4: and setting cwndGain to be 2 and pacingGain to be 1.25, and ending the end-to-end available bandwidth estimation algorithm based on Kalman filtering.
Step 7.5: state vector estimated by the above Kalman filtering algorithm
Figure BDA00036211645600000410
Obtaining real-time available bandwidth BW of end-to-end linkavlβ/α, and available bandwidth-delay product BDPavl＝BWavlminRtt, set Congestion Window cwnd ═ BDPavlcwndGain, step rate, pacingRate, BWavl*pacingGain。
Step 8: setting the sending interval of the data packet according to the stepping rate paging _ rate obtained in Step7
Figure BDA00036211645600000411
The time for sending the data packet next time can be obtained.
The invention has the beneficial effects that: the millimeter wave communication technology brings an air interface with higher bandwidth and lower time delay for a 5G network, and the high dynamic property of the millimeter wave communication technology also brings challenges for the design of a transmission protocol in the 5G network, particularly the design of a congestion control algorithm. Aiming at the characteristics of high bandwidth, low time delay and high dynamics in 5G millimeter wave communication, the congestion window adjustment strategy based on available bandwidth is adopted, the Kalman filtering algorithm is utilized to estimate the actual available bandwidth of the link, the stepping rate and the congestion window are set according to the estimated value, and the bandwidth utilization rate is effectively improved. Meanwhile, the rapid congestion sensing algorithm is used for detecting network congestion in time, the increase rate of a congestion window is controlled as early as possible before a network congestion point appears, the congestion is effectively relieved, the queue length is reduced, the round-trip delay is obviously reduced, and the high-efficiency transmission of data in a 5G network is realized. Simulation experiment results show that the invention realizes higher throughput and lower round-trip delay in three typical 5G scenes, has more obvious performance advantages particularly in scenes with high dynamic property, and is expected to realize high-efficiency data transmission in 5G application scenes such as ultra-high-definition video transmission, wireless broadband, unmanned driving and the like.
Drawings
FIG. 1 is a flow chart of a congestion control algorithm of the present invention;
FIG. 2 is a topology diagram of three 5G network scenarios of the present invention;
FIG. 3 is a diagram of SINR variation with time in three 5G network scenarios of the present invention;
FIG. 4 is a diagram of the Kalman-BR, BBR context-throughput variation of the present invention;
FIG. 5 is a diagram of the variation of the Kalman BR BBR in the scene-round trip delay according to the present invention;
FIG. 6 is a diagram of the Kalman-BR and BBR variation situation in scene two throughput of the present invention;
FIG. 7 is a diagram of the variation of the two round trip delays of the Kalman BR and BBR in the scene according to the present invention;
FIG. 8 is a diagram of the present invention Kalman-BR, BBR three-throughput variation in the scene;
FIG. 9 is a diagram of the variation of the Kalman BR and BBR in the scene with three round trip delays.
Detailed Description
The invention is further described with reference to the following drawings and detailed description.
Example 1: as shown in fig. 1, a kalman filter dynamic congestion control method for 5G millimeter wave communication specifically includes:
step 1: the sending end and the receiving end establish connection through TCP three-way handshake, and the sending end records the rtt value during handshake.
The initialization variables of the sending end are as follows: congestion flag ispipeefilled ═ false, initial stepping rate ═ init _ cwnd ═ 8/rtt, previous rate-delay trace point prePoint ═ 0, 0 in fast congestion sensing algorithm, initial state vector x in Kalman filtering0＝[0.505，-0.5]TCovariance matrix of process noise in Kalman filtering
Figure BDA0003621164560000051
Estimation error covariance matrix in Kalman filtering
Figure BDA0003621164560000052
Figure BDA0003621164560000053
Noise covariance matrix R ═ 1 in Kalman filtering]。
Step 2: when the sending end sends a data packet p each time, recording the time packet, delayed _ time sent by the data packet and the total data amount packet, delayed, confirmed so far at the head of the data packet p, and calculating and storing the sending rate send _ rate of the data packet p according to a formula (1):
Figure BDA0003621164560000054
where p.size denotes the size of packet p,
Figure BDA0003621164560000055
indicating the transmission interval of packet p.
Step 3: when the sending end receives an ACK packet, the average transmission rate delivery-rate of the data packet p confirmed by the ACK is calculated and stored according to the formula (2), the difference between the current system time and the sending time packet.
Figure BDA0003621164560000061
Where, delayed represents the size of a data packet successfully received currently, and the size of a data packet acknowledged by ACK is added every time a valid ACK is received, packet.
Step 4: the sending end selects the largest delivery _ rate value as the maximum transmission rate sample maxRate and selects the smallest rtt value as the smallest round-trip delay sample minRtt from the 10 sets of delivery _ rate and rtt values obtained recently.
Step 5: and the sending end judges whether the end-to-end link is fully utilized or not according to the congestion sign ispipeefilled, if the sign is true, the Step7 is entered, and if the sign is false, the Step6 is entered.
Step 6: the sender enters a start phase (STARTUP) and invokes a fast congestion sensing algorithm.
The Step6 comprises the following specific steps:
step6.1: entering a fast congestion sensing algorithm, and setting a current rate-delay track point (delivery _ rate, rtt) according to the average transmission rate and round-trip delay rtt of the data packet p obtained in Step 3.
Step6.2: and judging whether the previous track point is empty, namely whether the delivery _ rate or rtt in the track point is 0, if at least one item is 0, entering Step6.5, and if not, entering Step6.3.
Step6.3: calculating the slope k of a straight line formed by the previous track point and the origin point, namely pre-pointk＝k*curPoint.delivery_rate。
Step6.4: judging whether the round-trip time delay curPoint of the current track point is greater than or equal to rttk. If yes, setting the congestion flag ispipeefilled to true, indicating that the available bandwidth of the link is fully utilized, and entering step 6.5. If not, directly entering Step6.5.
Step6.5: and finishing the quick congestion sensing algorithm, and updating the parameter pretoint ═ curPoint. cwndGain is set at 2.89 and pacingGain at 2.89.
Step6.6: estimating maximum bandwidth delay product BDPmaxmaxRate minRtt. Setting Congestion Window cwnd ═ BDPmaxcwndGain, step rate pacingrte maxRate pacingGain.
Step 7: entering a stable probing available bandwidth stage (PROBE _ AB), and estimating the current available bandwidth by using an end-to-end available bandwidth estimation algorithm based on Kalman filtering.
The Step7 comprises the following specific steps:
step7.1: byThe send _ rate in Step72 and the receive _ rate in Step3 obtain a Kalman filtering measurement state zkSend _ rate/receive _ rate-1 and transition matrix H ═ send _ rate, 1 in measurement state and system state]。
Step7.2: the estimation process of the Kalman filtering algorithm comprises the following steps: using the initial state vector x in Step10And the formula (3) obtains the prior estimated value of the system state x
Figure BDA0003621164560000071
Obtaining a prior error covariance matrix of the prior estimate using equation (4)
Figure BDA0003621164560000072
Figure BDA0003621164560000073
Figure BDA0003621164560000074
Step7.3: the correction process of the Kalman filtering algorithm comprises the following steps: kalman gain K using equation (5)kThen using equation (6) to estimate a priori from the system state
Figure BDA0003621164560000075
And obtaining posterior estimated value from actual measured value zk
Figure BDA0003621164560000076
Finally, the error covariance matrix P of the posterior estimated value is updated by using the formula (7)kTo update in the next iteration
Figure BDA0003621164560000077
In formula (7), I is an identity matrix.
Figure BDA0003621164560000078
Figure BDA0003621164560000079
Figure BDA00036211645600000710
Step 7.4: and setting cwndGain to be 2 and pacingGain to be 1.25, and ending the end-to-end available bandwidth estimation algorithm based on Kalman filtering.
Step 7.5: state vector estimated by the above Kalman filtering algorithm
Figure BDA00036211645600000711
Obtaining real-time available bandwidth BW of end-to-end linkavlβ/α, and available bandwidth-delay product BDPavl＝BWavlminRtt, set Congestion Window cwnd ═ BDPavlcwndGain, step rate, pacingRate, BWavl*pacingGain。
Step 8: setting the sending interval of the data packet according to the stepping rate paging _ rate obtained in Step7
Figure BDA00036211645600000712
The time for sending the data packet next time can be obtained.
The method is realized on network simulation software NS-3 according to the steps and is named as Kalman-BR. And designing three network use scenes of static state, shielding and moving, and carrying out performance test on Kalman-BR and BBR.
The topology of the three scenarios is shown in fig. 2, wherein the 5G core network is located at the origin of the cartesian coordinate system and is connected to the server through a router. Fig. 3 shows SINR variation in three scenarios.
In scenario one, the distance between the ue and the base station is set to 25m, and both are performing line-of-sight communication with excellent signal quality. In scenario two, the distance between the user equipment and the base station is set to be 50m, and a building complete occlusion signal is set between the two, and the two perform non-line-of-sight communication. It can be seen from fig. 3 that the signal-to-noise ratio after occlusion is only half of scene one, and that some fluctuation of the signal occurs. In the third scenario, the user equipment moves continuously from 150m away from the base station to the base station at a constant speed of 5m/s, and keeps a longitudinal distance of 20 m from the base station. The signal fluctuations at this time are quite apparent from fig. 3.
Fig. 4 and 5 show the throughput and round-trip delay variation of Kalman-BR and BBR under scenario one, respectively. It can be seen that Kalman-BR has similar throughput performance to BBR in static scenarios, while the round trip delay is significantly lower than BBR. Specifically, the average throughput of Kalman-BR in a static scene is improved by about 3% compared with BBR, and the round-trip delay is reduced by about 6%. Compared with the mode that the BBR blindly increases the sending rate and then empties the queue to detect the available bandwidth, the Kalman-BR adopts the Kalman filtering algorithm and can gradually converge towards the current actual available bandwidth, and the bandwidth detection is more reasonable. However, the BBR blindly detects the bandwidth, which results in burstiness of data traffic, not only increases the queue, but also causes round-trip delay jitter.
Fig. 6 and 7 show the throughput and round-trip delay variation of Kalman-BR and BBR in scenario two, respectively. In the second scene, due to the obstruction of the obstacle, the signal quality is reduced, and certain jitter occurs. It can be seen from the figure that, at this time, the throughput of the Kalman-BR is not much different from that of the BBR, but the jitter of the throughput of the Kalman-BR is not severe as compared with that of the BBR, and a large change of a congestion window in the BBR causes a burst traffic in the network, resulting in a queuing phenomenon in a queue, so that the round-trip delay of the BBR is significantly higher than that of the Kalman-BR, as shown in fig. 7, the round-trip delay of the Kalman-BR is reduced by 5% compared with that of the BBR. It can be seen that the bandwidth detection mechanism of Kalman-BR is more stable and reasonable than the blind detection mechanism of BBR.
Fig. 8 and fig. 9 show the throughput and round-trip delay variation of Kalman-BR and BBR in scenario three, respectively. It can be seen that, when 0-20 seconds, a user gradually approaches a base station, the signal quality gradually becomes better, the throughput performance of Kalman-BR is improved by 41% compared with BBR, and the round-trip delay is reduced by 34% compared with BBR. When 20-40 seconds, a user is in an excellent signal coverage range of a base station, the throughput performance of Kalman-BR is improved by 12.9% compared with BBR, and the round-trip delay is reduced by 14.2% compared with BBR. At 40-60 seconds, the user gradually gets away from the base station, the signal quality becomes very poor, the average throughput of Kalman-BR is reduced by about 7.9% compared with BBR, but the round-trip delay is reduced by 20% compared with BBR. In general, compared with BBR, the throughput of Kalman-BR is improved by 14.8%, and the round-trip delay is reduced by 22.6%. Therefore, the Kalman-BR can better predict the bandwidth in a high-dynamic scene, and can quickly detect the optimal congestion control point when the signal quality changes and is about to be congested, so that the rate is reduced, a large amount of packet loss is avoided, and higher throughput and lower time delay are realized.
From the experimental results of the three experimental scenes, the Kalman-BR and BBR have no obvious difference in throughput in a relatively stable 5G network, but the round-trip delay performance of the Kalman-BR is better; in a mobile scene with high dynamic performance, the throughput and the round-trip delay of the Kalman-BR have obvious advantages compared with BBR (base band transceiver), and the Kalman-BR can be well adapted to the characteristics of high bandwidth, low delay and dynamic performance in 5G millimeter wave communication, so that the data transmission performance of a transmission layer in a 5G link is effectively improved.
While the present invention has been described in detail with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, and various changes can be made without departing from the spirit and scope of the present invention.

Claims (4)

1. A Kalman filtering dynamic congestion control method for 5G millimeter wave communication is characterized by comprising the following steps: at the beginning of TCP connection establishment, a sending end enters a starting stage, a congestion window cwnd and a stepping rate are exponentially increased, and a data packet is quickly sent to fill a link; meanwhile, a rapid congestion sensing algorithm is called to timely detect whether the link is fully utilized currently; when the link is detected to be fully utilized in the starting stage, entering a detection available bandwidth stage, calling an end-to-end available bandwidth estimation algorithm based on Kalman filtering to estimate the actual available bandwidth of the current link, and setting cwnd and payload according to the available bandwidth; and finally, the sending end controls the sending time of the next data packet according to the pacingrate.
2. The 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method according to claim 1, characterized by comprising the following specific steps:
step 1: the method comprises the steps that a sending end and a receiving end establish connection through TCP three-way handshake, and the sending end records an rtt value during handshake;
the initialization variables of the sending end are as follows: congestion flag ispipeefilled ═ false, initial stepping rate ═ init _ cwnd ═ 8/rtt, previous rate-delay trace point prePoint ═ 0, 0 in fast congestion sensing algorithm, initial state vector x in Kalman filtering0＝[0.505，-0.5]TCovariance matrix of process noise in Kalman filtering
Figure FDA0003621164550000011
Estimation error covariance matrix in Kalman filtering
Figure FDA0003621164550000012
Figure FDA0003621164550000013
Noise covariance matrix R ═ 1 in Kalman filtering]；
Step 2: when the sending end sends a data packet p each time, recording the time packet, delayed _ time sent by the data packet and the total data amount packet, delayed, confirmed so far at the head of the data packet p, and calculating and storing the sending rate send _ rate of the data packet p according to a formula (1):
Figure FDA0003621164550000014
where p.size denotes the size of packet p,
Figure FDA0003621164550000015
indicates the transmission interval of the data packet p;
step 3: when the sending end receives an ACK packet, calculating and storing the average transmission rate delivery _ rate of the data packet p confirmed by the ACK according to a formula (2), using the difference between the current system time and the sending time packet.
Figure FDA0003621164550000016
Where, delayed represents the size of the data packet successfully received currently, the size of the data packet acknowledged by ACK is added every time a valid ACK is received, packet, delayed represents the total amount of data acknowledged when the data packet p is transmitted, new represents the current time when ACK is received,
Figure FDA0003621164550000021
indicating the transmission time of the data packet p;
step 4: the sending end selects the largest delivery _ rate value as the maxRate of the maximum transmission rate sample and selects the smallest rtt value as the minRtt of the minimum round-trip delay sample from the 10 sets of delivery _ rate and rtt values obtained recently;
step 5: the sending end judges whether the end-to-end link is fully utilized or not according to the congestion sign ispipeefiled, if the sign is true, the Step7 is entered, and if the sign is false, the Step6 is entered;
step 6: the sending end enters a starting stage and calls a rapid congestion sensing algorithm;
step 7: entering a stable detection available bandwidth stage, and estimating the current available bandwidth by using an end-to-end available bandwidth estimation algorithm based on Kalman filtering;
step 8: setting the sending interval of the data packet according to the stepping rate _ rate obtained in Step7
Figure FDA0003621164550000022
The time for sending the data packet next time can be obtained.
3. The 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method according to claim 2, characterized in that Step6 comprises the following specific steps:
step6.1: entering a fast congestion sensing algorithm, and setting a current rate-delay track point (delivery _ rate, rtt) according to the average transmission rate delivery _ rate and round-trip delay rtt of the data packet p obtained in Step 3;
step6.2: judging whether the previous track point is empty, namely whether delivery _ rate or rtt in the track point is 0, if at least one item is 0, entering Step6.5, otherwise entering Step6.3;
step6.3: calculating the slope k of a straight line formed by the previous track point and the origin point, namely pre-pointk＝k*curPoint.delivery_rate；
Step6.4: judging whether the round-trip time delay curPoint of the current track point is greater than or equal to rttk(ii) a If yes, setting a congestion flag ispipeefilled true, indicating that the available bandwidth of the link is fully utilized, and entering Step6.5; if not, directly entering Step6.5;
step6.5: finishing the rapid congestion perception algorithm, and updating a parameter prePoint which is curPoint;
step6.6: estimating maximum bandwidth delay product BDPmaxmaxRate minRtt, set congestion window cwnd BDPmaxcwndGain, step rate pacingrte ═ maxRate ═ pacnggain.
4. The 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method according to claim 2, characterized in that Step7 comprises the following specific steps:
step7.1: from the send _ rate in Step72 and the receive _ rate in Step3, the Kalman filtering measurement state z is obtainedkSend _ rate/receive _ rate-1 and in measurement state and system stateIs given as [ send _ rate, 1 ═ of the transformation matrix H ═ send _ rate]；
Step7.2: the estimation process of the Kalman filtering algorithm comprises the following steps: using the initial state vector x in Step10And the formula (3) obtains the prior estimated value of the system state x
Figure FDA0003621164550000031
Obtaining a prior error covariance matrix of the prior estimate using equation (4)
Figure FDA0003621164550000032
Figure FDA0003621164550000033
Figure FDA0003621164550000034
Step7.3: the correction process of the Kalman filtering algorithm comprises the following steps: kalman gain K using equation (5)kThen using equation (6) to estimate a priori from the system state
Figure FDA0003621164550000035
And the actual measured value zkObtaining the posterior estimated value
Figure FDA0003621164550000036
Finally, the error covariance matrix P of the posterior estimated value is updated by using the formula (7)kTo update in the next iteration
Figure FDA0003621164550000037
In the formula (7), I is an identity matrix;
Figure FDA0003621164550000038
Figure FDA0003621164550000039
Figure FDA00036211645500000310
step 7.4: setting cwndGain to be 2, pacingGain to be 1.25, and ending an end-to-end available bandwidth estimation algorithm based on Kalman filtering;
step 7.5: state vector estimated by the above Kalman filtering algorithm
Figure FDA00036211645500000311
Obtaining real-time available bandwidth BW of end-to-end linkavlβ/α, and available bandwidth-delay product BDPavl＝BWavlminRtt, set Congestion Window cwnd ═ BDPavlcwndGain, step rate, pacingRate, BWavl*pacingGain。
CN202210463305.0A 2022-04-28 2022-04-28 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method Pending CN114710807A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210463305.0A CN114710807A (en) 2022-04-28 2022-04-28 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210463305.0A CN114710807A (en) 2022-04-28 2022-04-28 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method

Publications (1)

Publication Number Publication Date
CN114710807A true CN114710807A (en) 2022-07-05

Family

ID=82176900

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210463305.0A Pending CN114710807A (en) 2022-04-28 2022-04-28 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method

Country Status (1)

Country Link
CN (1) CN114710807A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116153079A (en) * 2023-04-17 2023-05-23 湖南工商大学 Road traffic flow prediction method and device based on vehicle track analysis
CN116684359A (en) * 2023-05-06 2023-09-01 铜仁职业技术学院 Network congestion control method, device and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105873096A (en) * 2016-03-24 2016-08-17 重庆邮电大学 Optimization method of efficient throughput capacity of multipath parallel transmission system
US20170353386A1 (en) * 2016-06-01 2017-12-07 Sichuan University Kind of congestion improvement method based on the quic protocol
CN110730469A (en) * 2019-09-03 2020-01-24 西安工业大学 Method for predicting bandwidth based on extended Kalman wireless network and congestion control thereof

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105873096A (en) * 2016-03-24 2016-08-17 重庆邮电大学 Optimization method of efficient throughput capacity of multipath parallel transmission system
US20170353386A1 (en) * 2016-06-01 2017-12-07 Sichuan University Kind of congestion improvement method based on the quic protocol
CN110730469A (en) * 2019-09-03 2020-01-24 西安工业大学 Method for predicting bandwidth based on extended Kalman wireless network and congestion control thereof

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHANG PING等: "Adaptive background updating algorithm for traffic congestion detection based on Kalman filtering and inter-frame centroid distanc", 《2015 IEEE ADVANCED INFORMATION TECHNOLOGY, ELECTRONIC AND AUTOMATION CONTROL CONFERENCE (IAEAC)》, 10 March 2016 (2016-03-10) *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116153079A (en) * 2023-04-17 2023-05-23 湖南工商大学 Road traffic flow prediction method and device based on vehicle track analysis
CN116153079B (en) * 2023-04-17 2023-06-23 湖南工商大学 Road traffic flow prediction method and device based on vehicle track analysis
CN116684359A (en) * 2023-05-06 2023-09-01 铜仁职业技术学院 Network congestion control method, device and storage medium
CN116684359B (en) * 2023-05-06 2024-02-06 铜仁职业技术学院 Network congestion control method, device and storage medium

Similar Documents

Publication Publication Date Title
Zhang et al. Will TCP work in mmWave 5G cellular networks?
CN111683393B (en) Adaptive congestion control method for dynamically adjusting gain coefficient
Kliazovich et al. Cross-layer congestion control in ad hoc wireless networks
CN114710807A (en) 5G millimeter wave communication-oriented Kalman filtering dynamic congestion control method
US7778164B2 (en) Communication apparatus and communication method for implementing a communication session based upon a congestion window
JP4189410B2 (en) Wireless communication apparatus and transmission control method
CN101854738B (en) Transmission control protocol method for satellite network
US8854992B2 (en) Artificial delay inflation and jitter reduction to improve TCP throughputs
JP4763772B2 (en) QoS measurement and monitoring in service-differentiated wireless networks
JP4708978B2 (en) Communication system, communication terminal, session relay device, and communication protocol realizing high throughput
Wang et al. TCP with sender-side intelligence to handle dynamic, large, leaky pipes
US20100011270A1 (en) Communication system, communication device, and communication method
US8144732B2 (en) Method for boosting downlink transmission to mobile station and system utilizing the same
US20120147840A1 (en) Method for boosting downlink transmission to mobile station and system utilizing the same
CN114422443B (en) Satellite network TCP congestion control method based on bandwidth estimation and congestion prediction
CN110572333A (en) transmission rate adjusting method and congestion control algorithm in bandwidth detection stage
US20180176136A1 (en) TCP Bufferbloat Resolution
WO2016065676A1 (en) Optimization method and system based on throughput rate and packet loss control cwnd
US11785442B2 (en) Data transport network protocol based on real time transport network congestion conditions
EP1513281A1 (en) A method and apparatus for link adaptation for packet transmission
Abdullah Enhancing the TCP Newreno Fast RecoveryAlgorithm on 5G Networks
KR101398589B1 (en) Apparatus and method for tcp performance enhancement in wireless network
KR20150125471A (en) METHOD AND APPARATUS FOR CONTROLLING CONGESTION IN A WIRELESS NETWORK USING Transmission Control Protocol
KR100772535B1 (en) Method for controlling congestion of node consisting ad-hoc network and apparatus thereof
CN113098732B (en) Method for calculating transmission time of effective load

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination