CN104936224B - Energy-efficient offloading method with dual-band AP in WLAN - Google Patents

Abstract

The invention discloses an energy-efficiency distribution method with dual-band APs in a WLAN, which mainly solves the problems of large time delay and power consumption and low throughput in wireless network access point equipment. The implementation plan is: 1. Use the single-frequency conversion algorithm in the AP device to select the optimal number of frequency bands. This algorithm is based on online sampling. When the single-frequency band can meet the service quality requirements of the network service, only the single-frequency band is used to transmit services. When not satisfied, open the next frequency band; 2. Determine the number of frequency bands by the single-frequency conversion algorithm, and use the funnel to buffer the business packets that need to be sent before sending the business packets. When the switching threshold of the funnel is exceeded, all will be buffered The frames are sent to the frequency band with a shorter transmit buffer queue. The invention improves the energy efficiency of users, reduces energy consumption for production equipment manufacturers, and can be used for wireless local area network WLAN to select frequency bands and network layer flow control in dual frequency bands.

Description

Energy-efficient offloading method with dual-band AP in WLAN

technical field

The invention belongs to the technical field of communication, and specifically relates to an energy-efficient flow distribution method of a wireless access point AP, which can be used for wireless local area network WLAN to select frequency bands in dual frequency bands and control network layer flow.

Background technique

In a WLAN, after the arrival of services, the problem of flow control must be solved. There are usually several distribution methods such as random distribution, redundant distribution, water injection distribution and frame aggregation distribution. However, with the explosive growth of mobile application business, the energy consumption of terminal equipment is increasing, which leads to higher and higher requirements of customers for technical specifications of mobile equipment. Moreover, the multi-band wireless local area network WLAN adopts the 802.11n protocol. Compared with the traditional AP equipment, the power consumption of the new equipment is increased because it is equipped with multiple ports, which increases the fixed consumption of circuit units. However, the above-mentioned traditional single distribution method is still Can not better solve the energy consumption problem. Therefore, better solutions must be proposed to deal with energy consumption.

In the wireless local area network, the configuration of the wireless access point AP has two frequency bands: 2.4GHZ and 5.8GHZ. During service transmission, flow control is required, and different frequency bands will be selected for different distribution methods. With the rapid development of wireless local area networks, the network has higher and higher requirements on throughput, energy efficiency and delay. However, the above-mentioned traditional shunting methods all have different defects:

Random allocation: Regardless of the actual situation of each frequency band, the service packets to be sent are randomly allocated to a certain frequency band, and the service allocation is performed without purpose, and its performance is naturally poor.

Redundant allocation: first consider the 2.4GHz frequency band, and assign service packages to the 2.4GHz frequency band first. Only when the 2.4GHz frequency band reaches saturation, will the business be allocated to the 5.8GHz frequency band. When the business load is light, only the 2.4GHz frequency band In the working state, the 5.8GHz frequency band is idle most of the time without considering the transmission capabilities of different frequency bands, so it cannot achieve good performance.

Water injection allocation: Before the service is sent to the terminal device, the AP device checks the transmission buffer queue working in the 2.4GHz and 5.8GHz frequency bands, and then sends the new service to the transmission buffer queue with the shortest queue length to achieve better performance. However, every time a service packet is sent, an 802.11 message PLCP Preamble and PLCP Header need to be sent, which increases the transmission overhead, and the delay cannot achieve good performance.

Frame Aggregation: By sending several service packets at one time and buffering the service packets before distributing them to different frequency bands, only one 802.11 message PLCP Preamble and PLCP Header needs to be sent, which reduces the overhead and thus the response time. The time is relatively small, and excessive delay is avoided, thereby improving system throughput. However, since the sending buffer queue lengths of different frequency bands are not considered when sending service packets, services cannot be transmitted more effectively.

Contents of the invention

The purpose of the present invention is to address the shortcomings of the above prior art, and propose an energy-efficient offloading method with dual-band APs in WLAN, so as to reduce transmission delay, improve throughput, and realize more effective service transmission.

The technical solution to realize the object of the present invention is: combine the existing water injection distribution and frame aggregation, first use the frame aggregation to buffer the incoming business packets, wait until the cached business volume reaches the upper limit of the frame aggregation buffer, and then use the water injection distribution to cache The service packets are sent to the frequency band with the shortest input buffer queue. Its implementation steps include the following:

(1) Initialize the current radio frequency configuration time, service arrival rate S and service throughput G, the AP node starts to send service packets, and the adjacent nodes receive service packets;

(2) Set the threshold value T _Thr of the sampling interval as a constant, and the change _{σ S} of the service arrival rate S is also a constant. Operation: if one of T _L ≥T _Thr or |S±S'|≥2σ _S is satisfied, then perform step (3), otherwise, perform step (5), where S' represents the average value of the service arrival rate ;

(3) The AP node returns to the single-band work, calculates the throughput G _S and the average power loss P ₁ of the current single-band work, and predicts the throughput G _d and the average power loss P ₂ of the second frequency band when switching to the dual-band;

(4) Record the current sending time T and the last sampling deadline T _E , and judge whether the sampling duration overflows:

If T _E < T _Thr is satisfied, the sampling duration does not overflow, and then step (3) is performed;

If T _E ≥ T _Thr is satisfied, the sampling duration overflows. At this time, if Then execute step (5), if not satisfied, then the AP node continues to work in the single frequency band, and executes step (8), wherein the fixed consumption of other components of the node is represented by a constant P _c ;

(5) The AP node returns to the dual-band work to process the incoming service package, that is, put the incoming service package in a pre-configured funnel, the funnel is equipped with a piston, and the piston is used to block the funnel before the funnel is not fully loaded;

(6) With the continuous arrival of business packages, before the funnel is not fully loaded, the weight of the funnel is constantly updated and the time to open the piston is rescheduled;

(7) Determine whether the weight in the funnel exceeds the length of the sending buffer queue:

If the weight in the funnel exceeds the length of the sending buffer queue, then all business packets are sent to the frequency band with the shortest length of the sending queue, and step (8) is performed;

If the weight in the funnel does not exceed the length of the sending buffer queue, continue to aggregate service packets and return to step (6);

(8) The AP node sends the service packet, and the receiving end receives the service packet.

Compared with the traditional shunting method, the present invention combines the dual advantages of water injection distribution and frame aggregation, and has the following effects:

1) Avoiding excessive time delay: when the service package arrives, the present invention buffers the service package before distributing it to different frequency bands, performs frame aggregation, and then dynamically updates the weight of the cached package. If the condition for opening the piston is met, then Send all aggregated frames to the frequency band with the shortest sending buffer queue, so that the aggregated frame only needs one header, which reduces the PLCP preamble and header overhead required for sending, shortens the sending time, and avoids excessive delay. The cache time is relatively small, which also avoids excessive delay, and then uses water injection to allocate and send cached service packets to further shorten the sending time and avoid excessive delay;

2) Improved energy efficiency: when the business load comes, the present invention first performs frame aggregation, and after the switch threshold arrives, sends the aggregated service packet, which reduces transmission loss, thereby reducing energy consumption and improving energy efficiency, while The service is allocated and distributed by water injection, which shortens the sending time, improves the throughput, and also improves the energy efficiency, wherein, energy efficiency=throughput/energy consumption.

Description of drawings:

Fig. 1 is the realization flowchart of the present invention;

Fig. 2 is to use the present invention to the throughput performance simulation result figure of four sorting strategies in different transmission rates;

Fig. 3 is to use the present invention to the energy efficiency performance simulation result figure of four shunting strategies in different transmission rates;

Fig. 4 is a graph showing the delay performance simulation results of four offloading strategies in different transmission rates by using the present invention.

Detailed ways

The technical solutions and effects of the present invention will be further described below with reference to the accompanying drawings.

In WLAN, the port of the wireless access point AP is configured with two frequency bands: 2.4GHZ and 5.8GHZ. First, the energy-saving method of the dual-band AP is designed, that is, the single- and dual-band conversion algorithm. This algorithm is based on online sampling and real-time network Performance statistics and evaluation, self-adaptive selection of the most energy-efficient port configuration, when the single frequency band can meet the service quality requirements of the network business, only use the single frequency band to transmit the business, and when it cannot meet the requirements, use the dual frequency band to split the traffic at the same time, On this basis, the present invention proposes a strategy of combining frame aggregation and water injection allocation for the frame aggregation feature in the 802.11n standard. More frames; when the weight of a buffered frame exceeds the switch threshold, it is inserted into a band with a shorter transmit buffer queue.

With reference to Fig. 1, the concrete realization steps of the present invention are as follows:

Step 1: Initialize network parameters:

The time to initialize the current radio frequency configuration is 0, the service arrival rate S and the service throughput G are both 0, the AP node starts to send service packets, and the adjacent nodes receive service packets.

Step 2: Select the frequency band by switching between single and dual bands.

Based on online sampling, the present invention performs statistics and evaluation of network performance in real time, and self-adaptively selects the most energy-efficient port configuration. The specific implementation steps are as follows:

2.1) Judging whether one of T _L ≥ T _Thr or |S±S'|≥2σ _S is satisfied, if one is satisfied, then execute step 2.2, and return to single-band work at this time, otherwise, execute step 3, and return to Dual-band work, where the sampling interval threshold value T _Thr is a constant, the change σ _S of the service arrival rate S is also a constant, T _L represents the time interval from the last sampling process, and S' represents the average value of the service arrival rate value;

2.2) The AP node returns to the single-band work, calculates the throughput G _S and the average power loss P ₁ of the current single-band work, and predicts the throughput G _d and the average power loss P ₂ of the second frequency band when switching to the dual-band, where, The current single-band throughput G _S and average power loss P ₁ are directly calculated according to the real-time performance of the network; the dual-band throughput G _d and the average power loss P ₂ of the second frequency band are obtained through the following predictions:

Use the service arrival rate S to estimate the throughput G _d when the dual-band is enabled, that is, use S≥G _d to evaluate the conditions for enabling the dual-band, according to the conditions for enabling the dual-band: The conditions for launching dual-band are:

At this point, turn on the second frequency band for shunting;

P ₂ according to the formula Estimate, wherein, P _TX represents the average power consumption when the current frequency band transmits business, and P _I represents the average power consumption when the current frequency band is idle;

2.3) Record the current sending time T and the last sampling deadline T _E , and judge whether the sampling duration overflows:

If T _E < T _Thr is satisfied, the sampling duration has not overflowed, and step 2.2 is performed;

If T _E ≥ T _Thr is satisfied, the sampling duration overflows. At this time, if Execute step 3, if it is not satisfied, the AP node continues to work in the single frequency band and sends the service packet, the receiving end receives the service packet, and the step execution ends.

Step 3: aggregate the sent service packets.

The AP node returns to the dual-band work to process the service packets that need to be sent, that is, before the service packets are sent to different frequency bands, they need to be buffered to aggregate more frames; when the weight of the cached packets exceeds the switch threshold, the packets to be sent The packet block is inserted into the queue with a shorter sending buffer queue length, and the specific implementation steps are as follows:

3.1) Put the incoming business package in a pre-configured hopper, the hopper is equipped with a piston, and the hopper is blocked by the piston before the hopper is fully loaded;

3.2) With the continuous arrival of business packages, before the funnel is not fully loaded, the weight of the funnel is constantly updated and the time to open the piston is rescheduled. The specific method of updating the weight of the funnel is calculated according to the following formula:

Among them, W represents the updated funnel weight, n represents the number of cached service packages, i represents the i-th arriving service package, T _i represents the time that the i-th arriving service package stays in the cache, and Upper_Bound represents frame aggregation upper bound.

Step 4: Send the aggregated service package.

If the weight in the funnel exceeds the length of the sending buffer queue, all service packets are sent to the frequency band with the shortest sending queue length, the AP node sends the service packet, the receiving end receives the service packet, and the step execution ends;

If the weight in the funnel does not exceed the length of the sending buffer queue, continue to aggregate service packets and return to step 3.

Effect of the present invention can be further described by following simulation:

1. Simulation conditions:

Based on the above-mentioned embodiment, for a dual-band AP in a WLAN, let one of the frequency bands be at 5.8 GHz with a bandwidth of 40 MHz, and the other frequency band be at 2.4 GHz with a bandwidth of 20 MHz. The dual-bands are configured with service loads, and the service transmission rate is sequentially from 100 to 200 Mbps. Incremental. Before the service package is sent to the end user, configure the funnel to aggregate the frames to be sent. When the conditions for opening the funnel are met, the service package is sent according to the water injection allocation.

2. Simulation content and result analysis:

Simulation 1: According to the above simulation conditions, the throughput performance of the four distribution schemes of random allocation, redundant allocation, water injection allocation and the present invention are respectively counted, and the results are shown in FIG. 2 .

It can be seen from Figure 2 that as the service transmission rate continues to increase, the performance of random allocation is the worst, followed by redundant allocation, followed by water injection allocation, and the throughput of the proposed scheme in the present invention is the largest, because:

Random allocation: Regardless of the actual situation of each frequency band, the service packets to be sent are randomly allocated to a certain frequency band, and the service allocation is performed without purpose, and its throughput performance is naturally poor;

Redundant allocation: first consider the 2.4GHz frequency band, and assign service packages to the 2.4GHz frequency band first. Only when the 2.4GHz frequency band reaches saturation, will the business be allocated to the 5.8GHz frequency band. When the business load is light, only the 2.4GHz frequency band In the working state, the 5.8GHz frequency band is idle most of the time without considering the transmission capabilities of different frequency bands, so the throughput cannot achieve good performance;

Water injection allocation: Before the service packet is sent to the terminal device, the AP device checks the transmission buffer queue working in the 2.4GHz and 5.8GHz frequency bands, and then sends the new service to the transmission buffer queue with the shortest queue length, but does not consider the service packet Frame aggregation, throughput does not reach very good performance;

The present invention combines the dual advantages of frame aggregation and water injection allocation, not only considers the transmission capabilities of different frequency bands, but also performs frame aggregation before service packet transmission, so the throughput is better than the above-mentioned traditional method.

Simulation 2: According to the above simulation conditions, the energy efficiency performance of the four diversion distribution schemes of random allocation, redundant allocation, water injection allocation and the present invention were counted respectively, and the results are shown in Figure 3:

It can be seen from Fig. 3 that as the service transmission rate continues to increase, the performance of random allocation is the worst, followed by redundant allocation, and water injection allocation again, and the energy efficiency performance of the present invention is the best, wherein, energy efficiency=throughput/energy consumption, This is because:

Random allocation: Regardless of the actual situation of each frequency band, the service packets to be sent are randomly allocated to a certain frequency band, and the business allocation is performed without purpose, and its energy efficiency is naturally poor;

Redundant allocation: first consider the 2.4GHz frequency band, and assign service packages to the 2.4GHz frequency band first. Only when the 2.4GHz frequency band reaches saturation, will the business be allocated to the 5.8GHz frequency band. When the business load is light, only the 2.4GHz frequency band In the working state, the 5.8GHz frequency band is idle most of the time without considering the transmission capabilities of different frequency bands, the throughput cannot achieve good performance, and the idle state consumes a lot of energy, so the energy efficiency performance is not optimal;

Water injection allocation: Before the service is sent to the terminal device, the AP device checks the transmission buffer queue working in the 2.4GHz and 5.8GHz frequency bands, and then sends the new service to the transmission buffer queue with the shortest queue length, but does not consider the frame of the service package Aggregation, the throughput cannot achieve good performance, so the energy efficiency performance is not optimal;

The present invention combines the dual advantages of frame aggregation and water injection allocation, not only considers the transmission capabilities of different frequency bands, but also performs frame aggregation before transmission, and the throughput is better than the above-mentioned traditional methods, so the energy efficiency is also better than the above-mentioned traditional solutions.

Simulation 3: According to the above simulation conditions, the time delay performance of the four distribution schemes of random allocation, redundant allocation, water injection allocation and the present invention are respectively counted, and the results are shown in FIG. 4 .

It can be seen from Fig. 3 that among the time delays of the four distribution schemes, the time delay performance of the scheme proposed by the present invention is the best, because:

Redundancy scheme: Due to the heavy business load, the dual-band buffer queue is fully loaded most of the time, which exceeds the buffer length of the send queue by 64. In this dynamic full load state, only a small part of the time the dual-band send buffer queue length is less than 64. The dual-band buffer queue does not change much, so the delay change is not obvious;

Random allocation: Regardless of the actual situation of each frequency band, the service packets to be sent are randomly allocated to a certain frequency band, and the service allocation is performed without purpose, and the delay is naturally very poor;

Water injection allocation: Before the service is sent to the terminal device, the AP device checks the transmission buffer queue working in the 2.4GHz and 5.8GHz frequency bands, and then sends the new service to the transmission buffer queue with the shortest queue length, but does not consider the frame of the service package Aggregation cannot reduce the PLCP preamble and header overhead required for sending, and the delay performance is not optimal;

The present invention: before distributing service packets to different frequency bands, first buffer them, perform frame aggregation, and then dynamically update the weight of the cached packets, and if the conditions for opening the piston are met, all the aggregated frames are sent to the frequency band with the shortest transmission buffer queue, Such an aggregated frame only needs one header, which reduces the PLCP preamble and header overhead required for sending, shortens the sending time, and avoids excessive delay, and the buffer time is relatively small, which also avoids excessive delay. Then use water injection to allocate and send cached service packets to further shorten the sending time and avoid excessive delay;

When the business load reaches 180-200Mbps, the transmission queue is fully loaded at this time, and there is not much difference between random allocation, water injection allocation and the proposed scheme, so the delay performance is close, but redundant allocation produces a lot of packet loss. The number of statistical packets decreases, so it seems that the delay performance is very good.

The above description is only a specific example of the present invention. Obviously, for those skilled in the art, after understanding the content and principle of the present invention, they can make changes in form and details without departing from the principle and structure of the present invention. Various amendments and changes, but these amendments and changes based on the idea of the present invention are still within the protection scope of the claims of the present invention.

Claims (3)

1. have the energy-efficiency offloading method of dual-band AP in a kind of WLAN, comprise the steps: (1) Initialize the current radio frequency configuration time, service arrival rate S and service throughput G, the AP node starts to send service packets, and the adjacent nodes receive service packets; (2) Set the threshold value T _Thr of the sampling interval as a constant, and the change _{σ S} of the service arrival rate S is also a constant. Operation: if one of T _L ≥T _Thr or |S±S'|≥2σ _S is satisfied, then perform step (3), otherwise, perform step (5), where S' represents the average value of the service arrival rate ; (3) The AP node returns to the single-band work, calculates the throughput G _S and the average power loss P ₁ of the current single-band work, and predicts the throughput G _d and the average power loss P ₂ of the second frequency band when switching to the dual-band; (4) Record the current sending time T and the last sampling deadline T _E , and judge whether the sampling duration overflows: If T _E < T _Thr is satisfied, the sampling duration does not overflow, and then step (3) is performed; If T _E ≥ T _Thr is satisfied, the sampling duration overflows. At this time, if Then execute step (5), if not satisfied, then the AP node continues to work in the single frequency band, execute step (8); (5) The AP node returns to the dual-band work to process the incoming service package, that is, put the incoming service package in a pre-configured funnel, the funnel is equipped with a piston, and the piston is used to block the funnel before the funnel is not fully loaded; (6) With the continuous arrival of business packages, before the funnel is not fully loaded, the weight of the funnel is constantly updated and the time to open the piston is rescheduled; (7) Determine whether the weight in the funnel exceeds the length of the sending buffer queue: If the weight in the funnel exceeds the length of the sending buffer queue, then all business packets are sent to the frequency band with the shortest length of the sending queue, and step (8) is performed; If the weight in the funnel does not exceed the length of the sending buffer queue, continue to aggregate service packets and return to step (6); (8) The AP node sends the service packet, and the receiving end receives the service packet. 2. The method according to claim 1, wherein said step (3) is predicted to switch to the average power loss P ₂ of the throughput G _d of dual frequency bands and the second frequency band, calculated according to the following formula: Use the service arrival rate S to estimate the throughput G _d when the dual-band is enabled, that is, use S≥G _d to evaluate the conditions for enabling the dual-band, according to the conditions for enabling the dual-band: The conditions for launching dual-band are: At this point, turn on the second frequency band for shunting; P ₂ according to the formula Estimated, where P _TX represents the average power consumption when the current frequency band transmits services, and P _I represents the average power consumption when the current frequency band is idle. 3. method according to claim 1, wherein said step (6) constantly updates the weight of funnel, calculates according to following formula: Among them, W represents the updated funnel weight, n represents the number of cached service packages, i represents the i-th arriving service package, T _i represents the time that the i-th arriving service package stays in the cache, and Upper_Bound represents frame aggregation upper bound.