KR100668771B1 - Efficient scheduling method for guaranteeing qos in wireless network - Google Patents

Abstract

The present invention relates to a scheduling method for guaranteeing quality of service in a wireless network, wherein a master performing scheduling allocates a time slot to receive a quality control packet including scheduling information from a slave transmitting a burst of each flow. Receiving scheduling information about the burst of each flow, generating a scheduling list using only bursts capable of completing transmission within a transmission time limit based on the scheduling information included in the quality control packet, and assignable Allocating the time slots to the bursts of the real-time flows included in the scheduling list for the time slots; and non-real-times included in the scheduling lists for the remaining time slots after completing slot assignments for the bursts of the real-time flows. To burst of flow And a step of assigning a group remaining time slots. At this time, preferably, the scheduling list generating step selects bursts of each flow that cannot be completed within the transmission time limit included in the quality control packet, and notifies the slave transmitting the bursts to delete them. It may include a step. In the slot allocating step, time slots may be allocated to bursts of the real-time flow in order of remaining time until the transmission time limit included in the quality control packet, and up to the transmission time limit included in the quality control packet. When the remaining time is the same, time slots may be allocated by selecting bursts in order of decreasing queue length of the flow.

Wireless network, QoS, scheduler, prediction algorithm

Description

Efficient Scheduling Method for Guaranteeing Quality of Service in Wireless Networks {EFFICIENT SCHEDULING METHOD FOR GUARANTEEING QOS IN WIRELESS NETWORK}

1 is an exemplary diagram of a superframe configuration of IEEE 802.15.3, which is a wireless network.

2 is a flowchart of a scheduling method according to a preferred embodiment of the present invention.

The present invention relates to a scheduling method for a wireless network. More particularly, the present invention relates to an efficient scheduling method for guaranteeing a quality of service (QoS) in a wireless network with a significant change in background noise. will be.

In general, in order to guarantee QoS in a data network, a scheduling algorithm must satisfy the following factors.

Fairness of bandwidth allocation

-Defined end-to-end delay time

Efficient use of network resources

At this time, in order for the scheduler to efficiently use network resources, the scheduler needs to know information about flow states. However, since the scheduler of one node generally does not have direct access to the internal state information (eg queue length, data generation rate) of other nodes, the performance of the scheduler is determined by how quickly and accurately this information is obtained. It depends greatly.

In this regard, for example, a solution used in a wireless Asynchronous Transfer Mode (ATM) network is to piggyback internal state information into a data packet. That is, a method of including not only data but also queue length information and internal state information in a data packet is used.

However, as described above, conventional scheduling algorithms generally piggyback the queue length and deadline information of a packet and transmit the packet back to the scheduler. This internal information is included in the Media Access Control (MAC) header of every packet and transmitted, causing a significant load on the network. Moreover, this method sometimes has the problem of generating excessive packets to transmit internal information within a predetermined time.

Meanwhile, FIG. 1 illustrates a configuration of a super frame in another IEEE 802.15.3 network, which is a wireless network. As illustrated, the super frame includes a beacon period, a contention period, and a slot allocation period. Beacons are used to set time allocations and convey management information of the PAN network, and contention intervals are used to convey commands and / or asynchronous data. Slot allocation intervals are used for command, isochronous stream, and asynchronous data connections. Meanwhile, the contention section uses a carrier sense multiple access with collision avoidance (CSMA / CA) scheme for media access, and the slot allocation section uses a time division multiple access (TDMA) scheme for media access.

With regard to QoS, real-time media types required for inter-application communication include video, voice, text, and file, and various requirements such as bandwidth and delay limit depending on the method of compressing such media. The condition exists.

The IEEE 802.15.3 communication protocol is a network in which a master and a slave exist, and are a centralized network in which resources for enabling communication are determined and scheduled at the master. In this system, the slave asks the master for the bandwidth it needs and the master allocates the slave the length of the interval.For QoS, the slave informs the master with the required delay limit and bandwidth, and the master allocates the slot based on this. do.

That is, the master communicates information on resources requested by each slave to a beacon, and each slave transmits its data in the slot interval to which the requested resource is allocated after receiving the beacon information. At this time, the application of each station has a bandwidth and delay limit required according to the type of media to communicate with. Here, when the master makes a beacon and allocates each media to a slot, whether the request is sufficiently referred to is a measure of ensuring the quality of each media to be communicated.

As described above, since the wireless network having a high change in the surrounding environment uses a wireless medium having a high bit error rate (BER) as a transmission medium due to background noise, interference, and multipath fading, the scheduler ensures the quality of service. Much information must be exchanged frequently for scheduling. Therefore, when using the above-described conventional technique, the amount of information exchange for scheduling is too large, or an unnecessary burst is scheduled to be transmitted, which can significantly reduce the performance of the network.

SUMMARY OF THE INVENTION In order to solve the above problems, an object of the present invention is to propose a scheduling method suitable for a wireless network, and in particular, an object of the present invention is to allow scheduling to be performed using only bursts required for minimum information exchange.

In order to achieve the above object, according to a first aspect of the present invention, there is provided a scheduling method for guaranteeing the quality of service in a wireless network, (a) from a slave to which the master performing scheduling transmits a burst of each flow; Allocating time slots to receive a quality control packet including scheduling information and receiving scheduling information for the burst of each flow; and (b) transmitting timeout based on scheduling information included in the quality control packet. Generating a scheduling list with only bursts capable of completing the transmission within (c) assigning the time slots to bursts of real-time flows included in the scheduling list for assignable time slots, and (d) For the remaining time slots after completing slot assignment for bursts of real-time flows, Allocating the remaining time slots to bursts of non-real-time flows included in the scheduling list.

In this case, preferably, the scheduling information included in the quality control packet includes a transmission timeout for the burst of each flow, a number of packets having the same transmission timeout in the burst, and a queue length of each flow. Can be configured.

Also, the step (b) may include (e) selecting a burst of each flow that cannot be completed within the transmission time limit included in the quality control packet, and notifying the slave transmitting the burst to delete the burst. It may include. In the step (c), the time slots may be allocated to the bursts of the real-time flow in the order that the remaining time until the transmission time limit included in the quality control packet is short. More preferably, in the step (c), In the case where the remaining time until the transmission time limit included in the quality control packet is the same, time slots may be allocated by selecting bursts in order of decreasing queue length of the flow.

According to a second aspect of the present invention, there is provided a computer-readable recording medium having stored thereon a program for performing each step of the aforementioned scheduling method.

Hereinafter, preferred embodiments of the present invention will be described.

1. Structure and Transmission Timing of Quality Control Packet

First, the preferred embodiment of the present invention uses a newly defined quality control packet to reduce the scheduling load. The quality control packet contains the node's internal state information and is sent to the master node where the scheduler is located only when necessary. Further, in the preferred embodiment of the present invention, the method of indicating the transmission timeout is different from the prior art. That is, the length of the scheduling timeout (transmission timeout) can be greatly reduced by expressing the number of superframes from the time when the burst occurs instead of the floating point. Accordingly, the aforementioned quality control packet includes the following information.

Transmission timeout: The timeout for which scheduling should take place (expressed as the number of superframes from the time the packet occurred).

Number of packets: number of packets with the same transmission timeout in the burst

Queue length: queue length for each flow

Next, with regard to the transmission timing of the above-described quality control packet, a method for reducing the number of times the quality control packet is transmitted to the master will be described.

Even if internal information is piggybacked in the data packet as in the related art, related information needs to be transmitted in a form other than the data packet due to the on / off characteristic of the flow. For example, the scheduler allocates time slots for a flow when it recognizes that there is a burst to send, so if there is a distance between bursts within the flow, then information about subsequent bursts is not piggybacked into the previous burst. Do not. In such a case, a case may occur in which the scheduler does not recognize the existence of a subsequent burst and thus does not allocate a time slot. To prevent this problem, the slave should provide a way to notify the master where the scheduler is located that a new burst has been created.

Therefore, in the preferred embodiment of the present invention, the time slot for transmitting the above-described quality control packet is allocated only when the scheduler is needed, and the slave transmits information on the new burst to the master through this time slot.

2. Scheduling of Real-Time Flows

In scheduling a real-time flow, the scheduler selects the flow based on the first selection criterion for the remaining time until the above-described transmission time limit for the first burst of each flow.

The second selection criterion is the queue length of the flow. For example, when a plurality of flows having the same time remaining until the transmission time limit exist for the first burst of each flow according to the above-described first selection criterion, flows having short queue lengths of flows are scheduled first. According to the second selection criterion described above, the network resource usage efficiency can be optimized by first processing the most urgent flow in a heavy network, and later processing the spare.

2.1 Prediction Algorithms for Real-Time Flows

In a network environment under heavy load, some flows may not be transmitted immediately even though they are transmitted immediately, and may be discarded at the receiving node. Therefore, it is necessary to schedule the burst to be discarded in advance by the scheduler predicting the discard situation or the discard possibility by the receiving node in advance. To this end, in the preferred embodiment of the present invention, the scheduler uses a two-step prediction algorithm.

First, one-step prediction is made in the first phase of scheduling and is performed for all flows. In the first stage prediction, the scheduler determines bursts that cannot be transmitted before the transmission timeout, and notifies the slave transmitting the flow to which the burst belongs. As described above, the slave receiving the notification recognizes that the burst cannot be transmitted and discards the burst, and transmits the information about the subsequent burst in the flow to the master through the aforementioned quality control packet. Accordingly, the longer the burst length, the sooner it can be determined whether the burst is scheduled.

As described above, through the first stage prediction, the master retains only information of bursts that can be transmitted within a time limit, thereby enabling more efficient scheduling.

Subsequently, two-step prediction is performed after the scheduler selects a flow to transmit according to the above-described flow selection criteria. After the master selects the burst to be scheduled to determine the exact time that the burst will be sent, it once again confirms that the time meets the transmission timeout. In the second stage, the master has more accurate information about the scheduling situation, so more accurate prediction is possible than in the first stage. In the first stage, there is no accurate information about the future situation of the network, so the loose judgment is made, while in the second stage, more rigorous judgment is made based on the accurate information.

3. Scheduling of non-real time flows

In scheduling, non-real-time flows have a lower priority than real-time flows. If a time slot remains after scheduling bursts of all real-time flows, scheduling of non-real-time flows is made. At this time, the scheduling is performed based on the queue length information indicated in the above-described quality control packet, because the slave does not transmit the transmission timeout and the number of packets for the non-real-time flow.

On the other hand, in some cases, the transmission timeout may be elapsed before the non-real-time flow is scheduled and transmitted, and the problem is that the master that does not have the transmission timeout information for the flow does not know the fact. Therefore, to avoid the above situation, the preferred embodiment of the present invention uses a prediction algorithm for non-real-time flow.

3.1 Prediction Algorithms for Non-real-Time Flows

The prediction algorithm of the non-real-time flow is similar to the one-step prediction of the real-time flow, except that it is performed in the slave rather than the master. That is, the slave transmitting the non-real-time flow determines whether the packet located at the first of the flow can be scheduled within the time limit. If transmission is not possible within the time limit due to a change in queue length, the node notifies the master, preventing the master from unnecessary scheduling. This allows the master to have more accurate information on each flow and optimize the schedule.

4. Scheduler Operation

Regarding the operation of the scheduler, FIG. 2 shows step by step operation of the scheduler according to the preferred embodiment of the present invention.

Referring to FIG. 2, scheduling of a scheduler is started (S100), and the scheduler first determines whether to allocate a time slot for a quality control packet at every scheduling. In step S105, a time slot for a quality control packet is allocated to receive a quality control packet from a slave transmitting each flow.

Next, in step S110, the bursts that do not need to be scheduled beyond the transmission time limit are selected through the first-step prediction described above. The scheduler then notifies the slaves transmitting the flows to which these bursts belong, and allocates time slots to the slaves, giving them the opportunity to update information about the flows. That is, the slave discards the burst that has timed out and then sends the master information on the burst to be transmitted through a quality control packet.

Subsequent to step S110, a scheduling list is generated with only bursts that can keep the time limit (S115). Subsequently, in order to select and schedule bursts one by one in the scheduling list, in step S120, the scheduler determines the amount of assignable time slots remaining in the superframe.

If it is determined in step S120 that an assignable time slot exists, it is determined whether a burst of the real-time flow exists in the above-described scheduling list so as to process the slot allocation for the burst of the real-time flow in advance (S125). If there is a burst of the real-time flow in step S125, the flow advances to step S130, in which, according to the above-described first and second selection criteria of the real-time flow, i) first selects the burst having the shortest remaining time until the transmission time limit. Ii) If the remaining time until the transmission time limit is the same, a burst having a short queue length of a flow is selected to allocate a time slot.

Subsequently, in step S135, it is checked whether the time slot allocated in step S130 satisfies the transmission timeout of the corresponding burst, according to the second stage prediction algorithm of the real-time flow described above.

If it is predicted that transmission is possible within the time limit in step S135, the scheduling for the corresponding burst is determined, and the corresponding burst is deleted from the aforementioned scheduling list (S140). Subsequently, returning to step S120 described above, if there are remaining time slots available for allocation, another burst is selected from the list, and this is repeated until there are no time slots available for allocation in the superframe.

On the other hand, if it is predicted that transmission is impossible within the time limit in step S135, the flow advances to step S145 to cancel scheduling for the burst and delete it from the scheduling list. Subsequently, the process returns to the aforementioned step S120 and the scheduling is repeated as described above.

As described above, the scheduling of the bursts of all the real-time flows having the time limit may be completed by the repetitive operations of steps S120 to S145.

Subsequently, the scheduling of the non-real-time flow will be described with reference to the above-described step S125.

If the time slot remains in the above-described step S120 after the scheduling of the burst of the real-time flow is completed, since the burst of the real-time flow no longer exists in the scheduling list in the above-described step S125, as shown in FIG. In step S150, the scheduling of bursts of the non-real-time flow is started.

In step S150, the scheduling of the bursts of the non-real-time flow selects the burst based only on the queue length as described above. That is, the burst of the flow having the shortest queue length is selected to allocate a time slot. Subsequently, in step S150, scheduling for the burst to which the time slot is allocated is determined, and it is deleted from the scheduling list (S155).

Subsequently, step S160 determines whether there is a burst of non-real-time flow in the aforementioned scheduling list, and if additional scheduling is required, returns to the above-described step S150 to repeat scheduling for the non-real-time flow. . If the burst of the non-real-time flow to be scheduled in step S160 no longer exists, the master terminates the scheduling and broadcasts the scheduling result in a beacon.

On the other hand, even when there is no time slot that can be allocated in the above-described step (S120), the master may terminate the scheduling, and broadcast the scheduling result in a beacon.

Although the preferred embodiment according to the present invention has been described above, this is merely exemplary and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the protection scope of the present invention should be defined by the following claims.

As described above, according to the present invention, it is possible to significantly reduce the amount and number of information exchanged for scheduling, and to prevent unnecessary bursts from being scheduled, thereby improving the ability of the wireless network to guarantee quality of service.

Claims (20)

A scheduling method for guaranteeing quality of service in a wireless network, (a) allocating a time slot to receive a quality control packet including scheduling information from a slave transmitting a burst of each flow by a master performing scheduling, and receiving scheduling information on the burst of each flow; (b) generating a scheduling list using only bursts capable of completing transmission within a transmission time limit based on the scheduling information included in the quality control packet; (c) assigning the time slots to bursts of real-time flows included in the scheduling list for assignable time slots; (d) assigning the remaining time slots to bursts of non-real-time flows included in the scheduling list for remaining time slots after completing slot assignments for bursts of real-time flows; Scheduling method of a wireless network comprising a. The method of claim 1, wherein the scheduling information included in the quality control packet is A transmission timeout for the burst of each flow, The number of packets with the same transmission timeout in the burst, Queue length of each flow Scheduling method of a wireless network to be configured. The method of claim 2, wherein step (b) (e) selecting a burst of each flow that cannot complete transmission within the transmission time limit included in the quality control packet, and notifying and deleting the burst that transmits the burst; The scheduling method of a wireless network comprising a. The method of claim 2, wherein step (c) And allocating time slots to bursts of the real-time flows in the order of remaining time until the transmission time limit included in the quality control packet. The method of claim 4, wherein step (c) When the remaining time until the transmission time limit included in the quality control packet is the same, selecting a burst in the order of shortest queue length of the flow and allocating time slots. The method according to any one of claims 1 to 5, wherein step (c) (f) confirming whether a time slot allocated to the burst of the real-time flow satisfies the transmission timeout of the burst. The scheduling method of a wireless network comprising a. The method of claim 6, wherein step (f) (g) if the time slot allocated to the burst of the real-time flow satisfies the transmission timeout of the burst, determining scheduling for the burst and deleting it from the scheduling list; (h) if the time slot assigned to the burst of the real-time flow does not satisfy the transmission timeout of the burst, canceling the scheduling for the burst and deleting from the scheduling list. The scheduling method of a wireless network comprising a. The method according to any one of claims 1 to 5, wherein step (d) And allocating time slots to bursts of the non-real-time flows in order of decreasing queue length of the flows. The method according to any one of claims 1 to 5, And the slave transmits a quality control packet to the master only for the burst of each real-time flow capable of completing the transmission within the transmission timeout. The method according to any one of claims 1 to 5, (i) subsequent to step (d), broadcasting a scheduling result The scheduling method of a wireless network further comprising. A computer-readable recording medium storing a scheduling program for guaranteeing a quality of service in a wireless network, the program comprising: (a) receiving a scheduling information for the burst of each flow by allocating a time slot to receive a quality control packet including scheduling information from a slave transmitting a burst of each flow by a master performing scheduling; (bb) generating a scheduling list using only bursts capable of completing transmission within a transmission time limit based on the scheduling information included in the quality control packet; (cc) assigning the time slots to bursts of real-time flows included in the scheduling list for assignable time slots; (dd) allocating the remaining time slots to the bursts of the non-real-time flows included in the scheduling list for the remaining time slots after completing slot assignments for the bursts of the real-time flows; And a computer readable recording medium. 12. The method of claim 11, wherein the scheduling information included in the quality control packet is A transmission timeout for the burst of each flow, The number of packets with the same transmission timeout in the burst, Queue length of each flow And a computer readable recording medium. The method of claim 12, wherein the step (bb) (ee) selecting a burst of each flow that cannot be completed within the transmission time limit included in the quality control packet, and notifying and deleting the burst that transmits the burst; And a computer readable recording medium. The method of claim 12, wherein the (cc) step And allocating time slots to the bursts of the real-time flow in the order of the remaining time until the transmission time limit included in the quality control packet. The method of claim 14, wherein the (cc) step And allocating time slots by selecting bursts in order of shortest queue length of the flow, when the remaining time until the transmission time limit included in the quality control packet is the same. The method according to any one of claims 11 to 15, wherein (cc) (ff) checking whether the time slot allocated to the burst of the real-time flow satisfies the transmission timeout of the burst And a computer readable recording medium. The method of claim 16, wherein the (ff) step (gg) if the time slot allocated to the burst of the real-time flow satisfies the transmission timeout of the burst, determining scheduling for the burst and deleting it from the scheduling list; (hh) if the time slot allocated to the burst of the real-time flow does not satisfy the transmission timeout of the burst, canceling the scheduling for the burst and deleting from the scheduling list. And a computer readable recording medium. The method according to any one of claims 11 to 15, wherein the step (dd) And allocating time slots to bursts of the non-real-time flows in order of decreasing queue length of the flows. The method of claim 11, wherein And said slave transmits a quality control packet to said mast only for a burst of each real-time flow capable of completing the transmission within a transmission timeout. The method according to any one of claims 11 to 15, (ii) subsequent to step (dd), broadcasting a scheduling result Computer-readable recording medium further comprising.

Images