Material Detail

This video is a part of External contributions. The goal of active networking is to create communication networks that reposition static, low-level network operation into dynamic, differentiated, and adaptable behavior. This allows communication hardware to be more fully used given that its operation can be tailored to specific application requirements. This also enables a more flexible and survivable communication network. Active networking decouples the network protocol from its transport by allowing easy insertion of protocols on top of the transport layer. Active networking also minimizes requirements for global agreement; it does not require years of standards negotiation to introduce new protocols. Active networking enables on-the-fly experimentation given easy insertion of new protocols and network applications, thus enabling the rapid deployment of new services and applications. The mechanism for implementing an active network is to enable communication packets to carry network code as well as data. This code may be installed on the fly into low-level network devices as the packet flows throughout the network. Ultimately, the essence and fundamental uniqueness of active networking is the flexibility introduced by a tight integration of code and data in service of the communication network. Both code and data flow within, and change the operation of, the network. Legacy networks have focused on improving the flow of data based on the fundamentals of analyses such as Shannon's fundamental insights into entropy as well as analyses in support of moving bits, such as queuing theory. With the tighter integration of code and data in an active network, a broader view of information encompassing both code and data in the form of Kolmogorov complexity is required. In this form of analysis, there is a focus on code and data as a combined entity. Active packets in the Kolmogorov complexity framework may vary from static data, as in legacy networks, to pure executable code; as code, the packets act as small executable models of information. This presentation delves into some of these issues.

Disciplines:

• Science and Technology  / Computer Science