57 results about "Random early detection" patented technology

A method and apparatus for implementing a very efficient random early detection algorithm in the forwarding path of a network device. Under one embodiment of the method flows are associated with corresponding Weighted Random Early Detection (WRED) drop profile parameters, and a flow queue is allocated to each of multiple flows. Estimated drop probability values are repeatedly generated for the flow queues based on existing flow queue state data in combination with WRED drop profile parameters. In parallel, various packet forwarding operations are performed, including packet classification, which assigns a packet to a flow queue for enqueing. In conjunction with this, a determination is made to whether to enqueue the packet in the flow queue or drop it by comparing the estimated drop probability value for the flow queue with a random number that is generated in the forwarding path.

The invention is to use the ability of a switching fabric to set a congestion indicator bit in a segment if any queue through which the segment passes is filled above a lower threshold. The output linecard monitors the field of the congestion indicator bit as it receives segments from the switching fabric. The output linecard periodically calculates the ratio of segments having the congestion bit set to all segments routed to a particular port. The periodically calculated ratio is used as an input parameter to a Random Early Detection (RED) algorithm. The RED algorithm selects a packet for the output linecard to drop, by use of a random selection method. The destination computer then does not receive the packet. The random selection of packets to drop has the effect of helping to prevent undesirable network synchronization of transmission of replacement packets. With adaptive source computers, the network device then does not reach a congested state, thereby maintaining optimum throughput for the computer network. When an ATM switching fabric is used with ATM cells for the segments, then the Explicit Forward Congestion Indication Field (EFCI bit) of a data cell is used to mark a data cell which has passed through a queue which is filled above a lower threshold level. Data cells arriving at the output line card with the EFCI bit set are then counted, and this count is used in the periodic calculation of the ratio used as input to the RED algorithm.

A network device identifies a non-adaptive flow as follows. The network device drops packets on a random basis using a Random Early Detection (RED) algorithm. A classifier reads indicia of a selected flow from at least one field of a header of a packet received by the network device. The network device calculates a drop interval for packets of the selected flow dropped by the RED algorithm, in response to a time at which the packets were dropped. The network device then applies a statistical test to drop intervals of a plurality of flows in order to identify the non-adaptive flow.

A Random Early Detection (RED) policer in accordance with the invention does not tail-drop arriving packets as being non-conforming. For instance, because the RED policer uses a running estimate such as an exponential weighted moving average (EWMA), for example, it allows the policer to absorb traffic bursts. The policer uses randomization in choosing which packets to drop; with this method, the probability of dropping a packet from a particular sending node is roughly proportional to the node's bandwidth share, hence the packets of different flows are fairly dropped.

Systems and methods for per service differentiation for congestion avoidance through dropping packets based on service priority include receiving an ingress packet; responsive to no congestion, providing the ingress packet to a queue of one or more queues; and, responsive to congestion, during a congestion window, one of providing the ingress packet to the queue and dropping the packet based on a packet dropping capability and service priority of a service associated with the packet.

A method for controlling data traffic with random early detection and window size adjustments including performing random early detection on incoming data packets, calculating a simple moving average of packet dropping probabilities for the data packets as calculated when performing random early detection, decreasing an advertised window size if the simple moving average is greater than a probability target plus a tolerance factor, increasing the advertised window size if the simple moving average is less than the probability target minus a tolerance factor, and not adjusting the window size if the simple moving average is not greater than a probability target plus a tolerance factor and not less than a probability target minus a tolerance factor.

Various techniques for queue management based on random early detection (RED) are disclosed herein. In particular, a method for generating a drop probability for an incoming packet in a device having a queue to buffer packets between segments of a network is provided. The method comprises determining, upon receipt of an incoming packet, a size of the queue and determining an error based at least in part on a difference between the queue size and a threshold. The method further comprises determining a drop probability for the incoming packet based at least in part on the error and a constant gain factor. The constant gain factor may be based at least in part on a linearized second order dynamic model of the network.

A switching fabric sets a congestion indicator bit (EFCI bit) in an ATM cell if any queue through which the ATM cell passes is filled above a lower threshold. A Traffic Manager monitors the field of the EFCI bit as cells arrive at an output port of the switching fabric. The traffic manager periodically calculates the ratio of ATM cells having the EFCI congestion bit set to all ATM cells routed to a particular output port. The periodically calculated ratio is used as an input parameter to a Random Early Detection (RED) algorithm. The RED algorithm selects a cell for the switch fabric to drop, by use of a random selection method. The destination computer then does not receive the packet since cells forming part of the packet have been discarded.

The invention discloses a congestion controlling method which is applied to a network technique field and comprises: packet identifiers that are corresponding to the packet slicing which is discarded in the process of random early testing and processing are extracted; the packet slicing with the same packet identifier is discarded. Internetwork protocol identifiers can be taken as the packet identifiers or adding identifiers; internetwork protocol identifiers or adding identifiers which are corresponding to the packet slicing which is discarded in the process of random early testing and processing are extracted; the extracted identifiers are compared with identifiers of subsequent packet slicing and if the identifiers are the same, the packet slicing are discarded, and if not, are retransmitted continuously. A receiving end receives and reorganizes the received packet slicing and when the reorganization goes wrong, retransmission is requested. The invention also discloses a congestion controlling device. The utilization of the invention can discard the useless relevant packet slicing in advance, thus alleviating the load of network devices, reducing the resource utilization rate, saving bandwidth and further ensuring high effectively carrying out other sessions without discarding packets.

The invention discloses an AQM (Active Queue Management) system and an AQM method based on a generalized PID (Proportion Integration Differentiation) random early detection algorithm. The system comprises a BP neural network module, a generalized PID control module and a random early detection module, wherein the BP neural network can be used for finding out parameters under a certain optimal control law by self learning; the generalized PID control module is used for performing proportion, integration and differentiation operations on regulated own parameters and stabilizing the queue length under different network loads at a corresponding fixed value; the random early detection module is used for acquiring the fixed value as a maximum dropping probability and generating a dropping probability suitable for a current load by self-regulation; a controlled object gives a corresponding response according to the dropping probability, so that the queue length in a router can be stabilized at an expected value. Therefore, the utilization rate of network resources is increased and the average time delay of a network can be reduced.

In order to maintain a small, stable backlog at an internet router buffer, active queue management (AQM) algorithms drop packets probabilistically at the onset of congestion. This causes Transmission Control Protocol (TCP) data flow rates to reduce. However, wireless losses may be misinterpreted as congestive losses and induce spurious reductions in flow rates. A prior art AQM with random early detection (RED), fails to maintain a stable backlog under time-varying wireless losses. However, the present invention can resolve the problem and provide a robust tracking of the backlog to a preset reference level by applying a control-theoretic vehicle, Internal Model principle, to realize such tracking. An integral controller (IC) is used as an embodiment of the principle that is robust against time-varying wireless losses under various network scenarios.

The invention relates to an interval type congestion control method for routers based on changeable load, which overcomes the over sensitivity of the existing RED(Random Early Detection) method to parameter setting and network environment and improves the threshold value in an RED from a point to an interval so as to obtain a new interval type RED method, namely, IRED (Interval Random Early Detection) algorithm, which obtains more strong robustness and better control under the load changeable network environment. Compared with the RED method, the performance of the system can be obviously improved by applying the method of the invention to carry out congestion control for routers, which leads to the decrease of loss rate of data packets, improvement of throughput of a network system and enhancement of stability. The method of the invention is a convenient and practical interval type congestion control method for the routers based on changeable load with excellent performance.

Flow control is provided for a packet switch. An incoming packet is classified according to its priority based on predefined rules to produce a weight value based on the priority of the packet. A lifetime value associated with the packet is produced based on the weight value and the queue occupancy in a queue to which the packet is assigned. The packet is discarded if its associated lifetime value is less than or equal to a threshold value. Otherwise the packet and associated lifetime value are sent to the queue. The lifetime value is periodically changed and compared to a threshold value. The packet is removed from the queue when its lifetime value reaches the threshold value.

The invention discloses a random early detection method and a random early detection device for data flow management. The method comprises the following steps of: acquiring first occupation depth of data content according to first length of the data content of a data packet which is to be input to a data cache space of a flow management chip, and current occupation depth of the data cache space; acquiring second occupation depth of a descriptor according to second length of the descriptor of the data packet and the current occupation depth of a descriptor cache space; and detecting the data packet according to the first occupation depth and the second occupation depth. The method and the device avoid a great amount of continuous detection caused by overflow of the descriptor cache space and achieve a technical effect of improving network quality.

The invention discloses a packet loss method in a long term evolution (LTE) system and a long term evolution-advanced (LTE-A) system based on an RED (Random Early Detection) algorithm, relates to a packet loss method, and aims to solve the delay problem of long waiting of delay sensitive data flow in a radio link control (RLC) layer buffer of the LTE system and the LTE-A system caused by congestion and even buffer overflow. The packet loss is controlled and implemented by changing the set value of a discard timer in the LTE system and the LTE-A system and comparing an instantaneous queue length in the buffer with a minimal threshold Qmin and a maximal threshold Qmax. The randomness of a probability P ensures the randomness and fairness of the packet loss, and is favorable for the algorithm to avoid continuous packet loss. The method provided by the invention is used for implementing the packet loss in the LTE system and the LTE-A system.

The invention relates to a nomadic-application-oriented network congestion control method. The network congestion adaptive method provides a congestion adaptive control algorithm based on up semi-normal distribution-DARED (distributed adaptive random early detection). According to the DARED algorithm, the corresponding congestion control is performed by improving an original data packet discarding probability function curve, namely adjusting an existing linear discarding straight line of an RED (random early detection) algorithm to a smooth curve, and dividing into different network states according to the queue length. The DARED is characterized in that a real-time dynamic parameter Pmax is introduced on the basis of the traditional RED algorithm, the stability of a queue is kept and the controllability of network congestion is increased. Correspondingly, in the aspect of a nomadic application system, when applied to the aspect of a multimedia seamless migration system, the congestion control algorithm realizes the seamless migration strategy under different architectures, namely the migration strategy under a B/S (browser/server) architecture, the migration strategy under a P2P (peer-to-peer) architecture and the migration strategy under a C/S (client/server) architecture. By adopting the nomadic application-oriented network congestion control method, the continuity, the rapidity, the smoothness and the correctness of media migration can be effectively ensured.