1. A new scheduling algorithm (COMA)
The iSlip schedule algorithm, which is adopted extensively in high speed routers, has the highest performance comparing with other algorithms up till now. However it still has some shortcomings such as small matching numbers. Based on the iSlip algorithm, we have designed a new algorithm named COoperation Matching Algorithm (COMA). In order to test the performance of the new algorithm comparing with iSlip, we use OPNET to do the simulation, with the aggregation of self-similar traffic model as the traffic source.

We collect the data of various performance metrics such as fairness, average number of matched packets in one round and delay time for queuing by running simulations in the conditions of diverse traffic intensity and distribution. The results show that in general COMA performs better than iSlip.

2. A Traffic Conditiner Implementation in IP Networks supporting QoS
Traditional IP network can only provide all customers with Best-Effort (BE) services. All the traffic compete equally for network resources. With the development of new applications of Internet, such as voice, video and www, the desire of QoS becomes more and more strong. Recently IETF proposes the Differentiated Services architecture to provide scalable means to deliver IP QoS based on handling of traffic aggregates. Traffic Classification state is conveyed by means of IP-layer packet marking using the DS field. Packets are classified and marked to receive a particular PHB (per-hop behavior) on nodes along their path. Sophisticated classification and traffic conditioning, including marking, policing, and shaping operations, need only be implemented at network boundaries. In this work, we propose to implement the single rate three color marker (srTCM) and two rate three color marker (trTCM), which has been RFCs, at the edge router of an IP networks supporting Differentiated Services at the simulation platform of OPNET.

3. A multi-threshod Rendom Early Detection(RED) Implementation
Floyd propose to use RED to improve the performance of responsive traffic which usually use TCP as the transport protocol, and to reduce the traffic congestion at the interior router of core network. However, with the requierment to support Differentiated Services with multiple traffic class and drop precedence, we need to enhance the RED to support it. The enhancement of RED can be ERED, RIO, Multi-RED, or WRED, while the implementation of these algorithm on OPNET is similar, therefore, we choose the implement mRED, which requires to use multiple threshold precedently drop packets according to their PHB. In this research, we propose to use the standard acb_fifo model in OPNET as the basis to implement mRED. We only need one queue (no subqueue), since one of the advantage of mRED is that it does not need to implement other scheduling algorithms at the output port, such as WFQ.

4. Research of the algorithms in high speed routers
As we have known. algorithms such as RED and iSlip are implemented in high speed routers. But these algorithms include a lot of parameters. How to config all these parameters in order to improve the performance of the router is a problem unsolved. So, we use OPNET to simulate the internal structure of routers and adjust the parameters to see how the router works in different conditions of parameters. We implement 6 different services in the router(Best Effort, Assured Forward(4 level) and Expediated Forward described in the related RFCs). And the traffic source is the aggregation of self-similar traffic model. First we tested the performance of iSlip by setting different speedup value. Then we do some simulation about the relationship between the buffer size and the discard rate. Having all the above parameters set properly, we set out to adjust the parameters of RED(Random Early Drop), which is relatively difficult to set. Lastly, we get all the values fit for the high speed router. We have also tested the performance of SCFQ. We added some code in the workstation model and the router model to implement the algorithm of SCFQ. The realization of SCFQ make the bandwidth allocation of AF service fair.