This paper describes ongoing research work to build a network management system from autonomous software
agents, using the BT Zeus Agent Building Toolkit. The network specific to this work is a 6-Node ATM Testbed,
which is fully reconfigurable; however, it is intended that the management system is generic enough to be implemented
on different networks. Each node of the network hosts its own agent society; within which the role of each
agent has an useful analogy to the controlling of the layers of the ISO OSI reference model. Although the lower-layer
agent's actions must be tightly prescribed, it is intended that higher-layer agents will take on more strategic, longerterm
outlooks, and be more adaptive and autonomous. Agent societies are able to communicate, to allow their collaboration;
it is intended that this will only be at the peer-to-peer level. The long-term aim of this project is to implement
fully distributed control within a network, while maintaining the ability to take more holistically informed
decisions for network-wide and longer-term strategies.
[1]
Guy Pujolle,et al.
Performance Management Issues in ATM Networks
,
1994,
INDC.
[2]
Nicholas R. Jennings,et al.
Agent-Based Computing: Promise and Perils
,
1999,
IJCAI.
[3]
Jean C. Walrand,et al.
High-performance communication networks
,
1999
.
[4]
Aluizio F. R. Araújo,et al.
Agent-Oriented Routing in Telecommunications Networks
,
2001
.
[5]
Nadia Boukhatem,et al.
Cooperative congestion control schemes in ATM networks
,
1996
.
[6]
Pravin Varaiya,et al.
High-performance communication networks (2nd ed.)
,
1999
.
[7]
Zheng Wang,et al.
An Architecture for Differentiated Services
,
1998,
RFC.
[8]
Michael Wooldridge,et al.
Agent-based software engineering
,
1997,
IEE Proc. Softw. Eng..
[9]
Anees Shaikh,et al.
Evaluating the impact of stale link state on quality-of-service routing
,
2001,
TNET.