摘要:
A time-based buffering system buffers data based upon how long the data should be held in order to comply with a traffic shaping policy. The data's source or destination need not be considered in determining where to buffer the data. The time-based buffering system includes a collection of time-based queues, each of which has a different time to dequeue. The system controlling traffic shaping determines how long a particular piece of data should be buffered (a “traffic shaping delay”) until it can be put on the network. Then, based upon that length of time, the system chooses one of the time-based of queues in which to buffer the data. That chosen queue has a dequeuing time that matches the traffic shaping delay. After the chosen queue dequeues its contents (at the specified time), it assumes a new dequeing time and is free to buffer new data that must be delayed by a time matching the new dequeuing time.
摘要:
A technique is described for improving packet performance in an access network. The access network comprises a Head End and a plurality of nodes. The access network further includes at least one shared-access upstream channel used by the first plurality of nodes to communicate with the Head End. Propagation delay data associated with at least a portion of the plurality of nodes using the at least one upstream channel is obtained. The propagation delay data is then used to dynamically adjust the lookahead time value associated with the generating of MAP messages for the at least one upstream channel.
摘要:
Unsolicited grants are allocated a selected time interval for scheduling transmission of audio packets at a network endpoint. A network processing node switches from outputting unsolicited grants to polling for packet transmission requests when Voice Activity Detection (VAD) at the transmitting endpoint stops generating audio packets. The network processing node switches back to outputting unsolicited grants when the endpoint resumes generation of audio packets. The unsolicited grants include one or more additional grants within the selected time interval that flush out one or more audio packets that may already be queued for transmitting. These additional grants reduce the latency normally caused when Voice Activity Detection (VAD) stops and then restarts audio packet transmission.
摘要:
Methods, apparatus, and computer-readable media are disclosed for performing diagnostic tests of a communication link between a headend and a cable modem without having to assign an IP address to the cable modem. The diagnostic tests are performed at the MAC layer and test the state of the hardware between and including the cable modem termination system (CMTS) and a selected cable modem. By doing so, a network operator is able to more easily focus in on a potential problem in cable RF/MAC connectivity between the two components by first eliminating (or identifying) a problem with hardware before focusing on problem-solving at the software level, typically at the Network layer, at which point the cable modem is assigned an IP address.
摘要:
A method of using DOCSIS 1.1 features to allow the addition of ISPs and QOS levels to a single cable modem without having to modify the CMTS is described in the various figures. Instead of using the SID of a data packet to determine the VPN tag of a data packet (DOCSIS 1.0), a service flow is used to identify the appropriate tag. This is done using the DOCSIS 1.1 configuration file. By doing so, the need for creating additional sub-interfaces in the cable modem interface does not arise. Instead, the configuration is modified at the provisioning server, i.e., the DHCP/TFTP server.
摘要:
Methods, apparatus, and computer-readable media are disclosed for creating a virtual lookahead upstream receiver in a single physical upstream receiver in a CMTS, thereby avoiding having to dedicate an upstream receiver strictly for lookahead capability. A lookahead receiver is used to determine whether a potential alternative frequency is better than the frequency presently being used. A physical upstream receiver is assigned to operate under a set of operational parameters associated with a logical lookahead receiver during a particular time slot. The logical receiver receives upstream data from a selected test modem using an alternative upstream frequency. It is then determined whether the alternative upstream frequency is preferable over the frequency presently being used. If so, the physical receiver is configured to operate normally under the set of operational parameters associated with the logical receiver. At this stage, all modems in a particular group, including the selected modem, hop over to the alternative frequency. The physical receiver can be divided into any number of logical receivers.
摘要:
A technique is provided for dynamically adjusting modem back-off parameters in a cable modem network. The technique is based on the theoretical result that the ratio of the number of the successful band width requests (Ns) to the number of collisions (Nc) has a value (e-31) when the back-off values are correctly estimated. The present technique for dynamically adjusting modem back-off parameters utilizes this concept to correspondingly correct the back-off parameters, depending upon the value of the Nc/Ns ratio, so as to cause this ratio to converge to the desired ratio of Nc/Ns equal to approximately 0.7. If the ratio of Nc/Ns is greater than a first specified value, the back-off start and backoff-end parameters (broadcast by the CMTS to cable modems on the downstream channel) are increased. If the ratio of Nc/Ns is less than a second specified value, the back-off start and backoff-end parameters are decreased. The present technique provides for improved access delay and improved throughput performance in contention sub-channels of cable access networks, particularly those involving large round trip delays. Additionally, the technique of the present invention utilizes elementary CPU operations, making it a viable and cost-efficient solution which is easily implemented and allows for rapid execution in existing cable modem network systems. Furthermore, the technique of the present invention is able to track the number of contending cable modems in a network over a much larger range than previous techniques.
摘要:
The technique of the present invention provides a simple and efficient solution to the problem of supporting differentiated priority levels within a QoS service class within a packet-switched network. When a bandwidth request is received at the cable modem head end, the service ID of that particular cable modem is identified. From this service ID, the associated static priority value of the requesting modem's service class is determined. The grant scheduler at the CMTS maintains a single queuing structure to temporarily store all differentiated priority bandwidth requests associated with a particular class of service that are received from cable modems on a selected channel. The technique of the present invention implements a procedure to calculate a metric used in determining a queuing priority for each received bandwidth request so that a single priority queuing structure may be used for this purpose. The metric is calculated by subtracting a product of the static priority value from the arrival time value of an associated bandwidth request. Use of the static service class priority in the queuing priority metric helps the grant scheduler to prioritize bandwidth requests from high priority modems over requests from low priority modems in the same queuing structure. Use of the arrival time in the metric enables an implicit fairness feature in the traffic prioritization to prevent starvation of low priority traffic.
摘要:
According to specific embodiments of the invention, a technique is described for providing constant bit rate (CBR) service over shared access channels in a computer network. The technique of the present invention provides a plurality of linked lists of grant allocation elements at the head end of the network for implementing CBR service to requesting nodes on each shared access channel. Each grant allocation element corresponds to a respective unsolicited CBR grant allocated to a particular node on a selected channel. A CBR service request sent by a particular node to the head end is handled by an admission control engine within the head end. If admission control admits the CBR service request, it inserts one or more unsolicited grant allocation elements relating to the admitted request in at least one of the plurality of linked lists. Each linked list of grant allocation elements is used to generate a CBR portion of a respective, fixed size, grant allocation MAP message. Each grant allocation MAP message includes a first portion of N minislot allocations reserved primarily for CBR purposes, and a second portion of M minislot allocations reserved primarily for non-CBR purposes. Using the technique of the present invention voice traffic or other CBR applications may be supported on shared access channels with delay bound and zero jitter.
摘要:
An approach for managing the consumption of resources uses adaptive random sampling to decrease the collection of flow statistical data as the consumption of resources increases. When a packet is received from a network, a determination is made whether the packet belongs to an existing flow, for which flow statistical data is being collected, or to a new flow. If the packet belongs to an existing flow, then the flow statistical data for the existing flow is updated to reflect the packet. If the packet belongs to the new flow, then a sampling probability is used to determine whether the new flow is to be sampled. The sampling probability is determined, at least in part, upon a current usage of resources.