JP2005094761A - Differentiated scheduling method in downlink communication between rnc and node-bs in network - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the unnecessary use of a bandwidth in a wireless interface or a bottleneck link and lower a discard rate of packets. <P>SOLUTION: This method controls downlink communication between an RNC 1 and node-Bs 2 in a network, particularly in an RAN in a UMTS. At least one router 3 is installed between the RNC 1 and a Node-B 2 and transmits packets. A bottleneck link is established between the router 3 and each node B. A packet transmitted by the RNC is delayed before transmission by the RNC in association with a window request based on the priority. This control method provides the data transmission of high quality and has high efficiency when using the bottleneck link. For the RNC, virtual queues are provided to all bottleneck links and the scheduling of the virtual queue is differentiated by the priority of the transmitted packet. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a differentiated scheduling method in downlink communication between an RNC (Radio Network Controller) and a Node B (Node-Bs) in a network.

  The present invention relates to a method for providing downlink differentiated scheduling between RNC (Radio Network Controller) and Node B (Node-Bs) in a network. The network particularly refers to a RAN (Radio Access Network) in the UMTS (Universal Mobile Telecommunication System), in which a router is installed between the RNC and a single Node B for packet transmission. A bottleneck link is formed by connection between the nodes B. A packet transmitted by the RNC is delayed by the RNC in response to a window request in the Node B based on the priority order before transmission.

  The thing similar to this invention is a well-known thing, and is utilized by various methods based on each condition. In general, there are special requirements for communication between RNC and Node B in UMTS-RAN. These requests specify in particular that the elapsed time from when a packet is transmitted from the RNC to the opposite Node B must be within a defined time limit. Packets that do not satisfy these requirements, that is, packets that arrive after the maximum allowable time has elapsed are discarded. Suppose that the radio frame MAC protocol for the transmitted frame is processed directly before transmission from the RNC. For this reason, the reception window constraint condition starts counting from the time when the frame is transmitted. A specific feature of downlink communication is that the packet can be delayed before being sent from the RNC by a receive window request. No delay is possible after transmission from the RNC.

  The well-known standard queuing mechanism has drawbacks in the following typical cases. That is, when there is a bottleneck link between the last router and the Node B, the RNC can easily fill the corresponding buffer of the last router so that the bottleneck link can be easily used in the last mile link (Last Mile Link). Can be overloaded.

  If the buffer has a capacity that allows the latency to be longer than the receive window, when overloaded, the buffer is almost full and when all packets reach it, all packets have their Since the destination is reached after the time limit is reached, it is not accepted. In order to avoid this problem, the buffer must have a capacity such that the above situation is unlikely to occur.

  One well-known method of managing downlink communication between the RNC and Node B in the bottleneck link is called router priority scheduling. The priority level of the transmission packet is selected according to the type of protocol and each application, for example. Thus, flexible applications such as e-mail and ftp that can handle resources freely are generally given low priority. On the other hand, real-time applications such as teleconferencing and VoIP (Voice over IP) applications are given high priority.

It is possible to provide a packet having a severe delay condition with a lower delay than a packet having a mild delay condition by scheduling the packet by priority scheduling.
However, even in this case, there is a problem that a packet having a low priority is transmitted if there is no packet having a higher priority. This means that it is very likely that a low priority packet will arrive at the destination with a delay greater than the receive window during overload. As a result, these packets are discarded. This causes unnecessary consumption of bandwidth on the bottleneck link, similar to the radio interface.

A well-known method as an alternative to the above-described priority scheduling and which can provide delay guarantees for all packets is the WRR (weighted round robin) scheme.
In this method, emphasis is placed on high-priority transfer, and in the case of overload, the low-priority packets are discarded while satisfying delay requests for both high- and low-priority packets.

  The disadvantage of the WRR method is that it does not adequately handle packets with mild delay conditions. This is because such packets are discarded instead of delayed if they do not meet their deadline. Furthermore, in this approach, the radio interface is not used effectively because each slot of the missing packet remains unused.

  An object of the present invention is to provide a method of priority differentiation by scheduling in downlink communication in which unnecessary consumption of bandwidth in a radio interface or bottleneck link is reduced.

  In addition, a method for differentiating priorities by scheduling in downlink communication for reducing the packet discard rate is provided.

  The present invention focuses on designing and improving methods in the field related to the scheduling process of downlink communication between RNC and Node B in the network. Here, the known problems mentioned above or in the current state of the art are simply avoided. At the same time, when a bottleneck link exists, high-efficiency and high-quality data transfer is realized.

  The method of the present invention used in the above fields solves the above problems. This is due to the features of the present invention as set forth in claim 1. That is, a feature of the present invention is that a virtual queue is provided for each bottleneck link in the RNC, and scheduling of the virtual queue differentiates between priorities of transmission packets.

  Various reasons are recognized according to the present invention. Differentiating by scheduling at the router is not a good choice in order to give short delay times to packets with high priority. According to the present invention, an alternative to the differentiating scheduling of the router is to schedule the packet according to the priority of the packet at the RNC, the reception window constraint after the packet is transmitted from the RNC is counted It is further recognized that packets with low priority without a delay requirement can be delayed at the RNC without being threatened by the time limit. According to the present invention, it is proposed that for each bottleneck link transmitted by the RNC, various virtual queues are maintained in the RNC. All virtual queues in the RNC are controlled by adding a difference between delay priorities. The packet delay is given according to the packet request. For example, VoIP packets can be scheduled with a shorter delay time than best effort data packets. For example, the division between delay priorities is performed by priority scheduling. The virtual queue is emptied at certain intervals so that the bottleneck link buffer is not used beyond the limit determined by the window request.

  With the method according to the invention, it is recognized that scheduling prioritizes higher priority packets. Using information available at the RNC, the high quality data traffic flow is protected against the low quality data traffic flow. In other words, for example, by giving the highest priority to VoIP data higher than flexible data such as e-mail, the voice quality is protected even at the peak of data traffic.

  Furthermore, it ensures that the packet is not discarded. Therefore, the bottleneck link and the radio interface are utilized to the maximum extent. In addition, downlink routers do not require support for any kind of differentiated scheduling. Simply sending packets over these links can maximize the utilization of potentially low capacity links between the RNC and Node B, and the packets are limited in time, i.e., the reception window related limitations. Arrives before time. By ensuring that the RNC transmits all packets, the efficiency of use of the radio interface is maximized, so that if there are time slots allocated to the radio interface, the packets will not be dropped and will exceed their time limit. However, since the node B is reached with a short delay time, the slot of the radio interface does not become empty.

In a more preferred embodiment, the virtual queue is divided into two sub-queues: a sub-queue for high priority packets and another sub-queue for packets with low priority. The method for controlling the transfer of packets in these two sub-queues uses this differentiation. For example, a high priority class with strict delay requirements includes voice packets, and a low priority class without delay requirements includes email data. For packets with a high priority, a delay between the RNC and the Node B is guaranteed, especially under the defined maximum value D _HP . This maximum delay time is a parameter that can be set without the system.

  In this embodiment, the virtual queue is designed to be zero at a rate corresponding to the capacity of each bottleneck link. That is, the rate of zeroing the virtual queue is selected in such a way that the bottleneck link buffer is not filled more than the defined capacity (given by the receive window request) defined. This choice ensures that the buffer on the bottleneck link is always empty. That is, the queue delay time is equal to zero. For this reason, there is no longer a delay than the reception window permits.

To realize the processing for delaying the packets just high priority D _HP, it is possible to determine the capacity of the sub-queues of higher priority in the virtual queue. At this time, the size of the sub queue is defined by the product of D _HP and the capacity Cb of the bottleneck link, where Cb is a rate at which the virtual queue is emptied. When the subqueue is already full, all incoming high priority packets are discarded because they cannot meet the delay guarantee conditions for these packets.

  In order to avoid discarding packets with high priority and limit the amount of packets with high priority to be input to the network, input control for packets with high priority is performed.

  In the form of an algorithm for emptying the virtual queue, it is beneficial to have the packets wait to adjust the transmission interval between the two packets during a configurable period. Specifically, this period is defined by the ratio between the packet length l and the bottleneck link capacity Cb.

  Instead of an embodiment in which all packets are kept in the RNC internal buffer, an embodiment can be provided in which packets are sent to the network and the status of all buffers on the path from the RNC to the Node B is recorded. . To record the status of these buffers, the RNC checks if the packet transfer causes a delay greater than the receive window. Thereby, packet transfer can be scheduled.

  In terms of cost, which increases the complexity of the algorithm for scheduling packet transfers, such an embodiment is more sensitive to the data buffer in the RNC than the embodiment where all packets remain in the RNC's internal buffer. Shows the benefit of fewer memory requirements involved.

  Communication within the RAN may be an IP (Internet Protocol) based network. Alternatively, it can be assumed that the present invention is applied to transfer over a network using ATM.

  Advantageously, there are different possibilities for designing and developing the content of the present invention. On the other hand, the dependent claims of claim 1 should be further referred to the description of preferred embodiments of the invention with reference to the following drawings. With respect to the description of the preferred embodiment of the present invention with reference to the drawings, the design and development of the generally preferred content will be described in the drawings described.

  According to the present invention, unnecessary consumption of bandwidth in a wireless interface or a bottleneck link is reduced.

  In addition, the packet discard rate is reduced.

  FIG. 1 schematically shows a standard UMTS-RAN configuration. There is a router 3 for packet transfer between RNC 1 and Node B (Node-Bs) 2 (only one is shown). In the last mile link between the router 3 and the node B2, there is a bottleneck link 4. For simplicity, it is assumed that there is only one bottleneck link in the path from RNC 1 to Node B 2 and this bottleneck link 4 is used only by a single RNC.

  FIG. 2 shows the same network configuration as that shown in FIG. 1, and the same components have the same components. In the router 3, the scheduling of packets received from the RNC via priority scheduling (PS) is schematically shown.

  FIG. 3 also shows the same network configuration along with packet scheduling in the router 3, but is based on weighted round robin (WRR) rather than priority scheduling.

FIG. 4 schematically shows an example of the implementation of the method of the invention. The RNC 1 transmits a packet to the two routers 3, and each router 3 transfers the packet to the node B 2 via the bottleneck link 4. Each bottleneck router schedules packets in a FIFO manner, and the buffer capacity is such that even if the buffer is filled, the received packets are not delayed beyond the size of their receive window.
The RNC 1 includes two virtual queues 5, each of which corresponds to the bottleneck link 4. Each virtual queue 5 of the RNC 1 is scheduled in a manner that differentiates delay priorities. In the embodiment shown in the figure, differentiation is performed by priority scheduling (PS). The virtual queue 5 is emptied at a rate such that the bottleneck link 4 buffer is only filled to the limit set by the receive window request.

  First, a simple case where only two priorities are differentiated will be described. That is, there are two categories: a high priority category (eg, voice) that requires a strict delay requirement and a low priority category (eg, data) that does not require a delay requirement. Next, as shown in FIG. 4, it is beneficial to divide each virtual queue 5 into two sub-queues 6. Packets with high priority are sorted and stored in sub-queue 6, and packets with low priority are sorted and stored in the other sub-queue 6. Differentiation is performed by a method of managing transfer of packets in the two subqueues 6.

FIG. 5 is an enlarged view of the virtual queue 5 having two sub-queues 7 and 8, and the virtual queue 5 is processed so as to be differentiated between delay priorities by priority scheduling. The upper subqueue 7 holds packets with high priority.
The capacity of the sub-queue corresponding to the high priority in the virtual queue 5 is determined so that the packet having the high priority is not delayed beyond the maximum delay time D _HP . Therefore, sub-queue 7 of higher priority _is defined by the product of _{D HP} and _{C b.} C _b represents the capacity of the bottleneck link 4 and is also the rate at which the virtual queue 5 is emptied.

FIG. 6 is a flowchart of an algorithm for emptying the virtual queue 5. First, it is checked whether each packet to be transmitted is a high priority packet. If the packet is high priority, the packet is transmitted and waits for the next packet transmission by the time of l / C _b. “L” means the length of the packet.
If it is not a high priority packet, it is checked as a next step whether it is a low priority packet. If it is a low priority packet, it is transmitted and similarly waits for l / _Cb time until the next packet is transmitted. If it is not a low priority packet, wait for a new packet to arrive.

  FIG. 7 is another flow diagram of the algorithm for emptying the virtual queue 5. Unlike the example of FIG. 6, in the flow of FIG. 7, not all packets are held in the internal buffer of the RNC 1, but are transmitted to the alternative network. On the other hand, an example is shown in which the states of all buffers on the path from RNC 1 to Node B 2 are recorded.

First, the control algorithm checks whether there is a packet in the high priority sub-queue (HP queue) 7.
If the packet is transferred is used is information available in RNS1, packets of higher priority, the delay time is checked whether the condition that not greater than D _HP.
If this condition is met, a second check is performed to ascertain whether the packet arriving at Node B 2 has a delay smaller than its reception window.
If both conditions are satisfied, the packet is transferred. If only the first condition is satisfied, the packet transfer is delayed, otherwise the packet is discarded. Confirmation of these conditions and packet delay are executed in the processing portion of “delay time <D _HP ” in FIG.

  As soon as the high priority subqueue 7 becomes empty, the same processing is executed for the low priority subqueue 8. If the expected delay of the packet is less than the receive window, the packet is forwarded. Otherwise, the packet is delayed. Note that the low priority data packet is not discarded because it has no delay request.

  Finally, it should be noted that the above description of the embodiment is merely an example of a solution and should not be limited to the present embodiment.

FIG. 1 is a schematic diagram of a standard UMTS-RAN. FIG. 2 is a schematic diagram of router priority scheduling in UMTS-RAN. FIG. 3 is a schematic diagram of router WWR scheduling in UMTS-RAN. FIG. 4 is a schematic diagram of an embodiment of RNC scheduling according to the present invention. FIG. 5 is a schematic diagram of a virtual queue having two subqueues. FIG. 6 is a flowchart of the first embodiment of the algorithm for emptying the virtual queue. FIG. 7 is a flowchart of the second embodiment of the algorithm for emptying the virtual queue.

Explanation of symbols

1 RNC
2 Node B
3 Router 4 Bottleneck link 5 Virtual queue 6 Subqueue 7 High priority subqueue 8 Low priority subqueue

Claims (13)

In a network, particularly a RAN (Radio Access Network) in a UMTS (Universal Mobile Telecommunications System), a link between the RNC (Radio Network Controller) (1) and the Node B (Node-Bs) link (2) is down. A method,
At least one router (3) is installed between each RNC (1) and each Node B (2) for transmitting packets,
A bottleneck link is formed in the path between the Node B (2) and each router (3),
The packet transmitted by the RNC (1) is delayed according to the reception window request in the Node B (2) based on priority before transmission,
In the RNC (1), a virtual queue (5) is prepared for each bottleneck link (4),
A method for differentiating between the priorities of the transmitted packets by controlling the virtual queue (5). The method of claim 1, comprising:
The virtual queue (5) is divided into a sub-queue (7) for high-priority packets and a sub-queue (8) for low-priority packets . The method according to claim 1 or 2, wherein
A method, characterized in that the delay time for high priority packets is guaranteed to be shorter than a configurable maximum value D _HP . The method according to any one of claims 1 to 3,
Method according to claim 1, characterized in that the virtual queue (5) is emptied at a rate corresponding to the capacity Cb of the respective bottleneck link (4). The method of claim 4, comprising:
Higher the relative packet priority sub-queue (7), wherein the capacity size defined by product of the capacitance Cb of the D _HP and bottleneck link is determined. The method according to any one of claims 1 to 5, comprising:
A method wherein input control is performed on a high priority packet when a high priority packet enters the virtual queue (5). The method according to any one of claims 1 to 6, comprising:
A method characterized in that a transmission interval between two packets can be set. The method of claim 7, comprising:
The method is characterized in that the transmission interval is defined by a ratio between a packet length l and a capacity Cb of the bottleneck link (4). The method according to any one of claims 1 to 6, comprising:
By considering the status of the router buffers on all single or multiple paths from the RNC (1) to the corresponding Node B (2), the RNC (1) to the Node B (2 ) Transmitting the packet. A method according to any one of claims 1 to 9,
Method according to claim 1, characterized in that the virtual queue (5) is divided into two or more subqueues (7) with different priorities. A method according to any one of claims 1 to 10, comprising
In the case of soft-handover, the method considers all the states of the associated virtual queue (5). 12. The method according to any one of claims 1 to 11, comprising:
A method comprising communication in an IP (Internet Protocol) network.   The method according to any one of claims 1 to 11, wherein the method is communication in an ATM (Asynchronous Transfer Mode) network.

Images