All optical packet switching (AOPS) technology is essential to fully utilize the tremendous bandwidth provided by ad vanced optical communication techniques through forward ing packets in optical domain for the next generation net work. However, long packet headers and other complex op erations such as table lookup and packet header rewriting still have to be processed electronically for lack of cost effective optical processing techniques. This not only in creases system complexity but also limits packet forwarding speed due to opticalelectronicoptical conversion. Lots of work of improving optical processing techniques to realize AOPS is reported in the literature. Differently, this paper proposes a new networking structure to facilitate AOPS realization and support various existing networks through simplifying networking operations. This structure only re quires an AOPS node to process a short packet header to for ward packets across it with neither table lookup nor header rewriting. Furthermore, it moves high layer addressing is sues from packet forwarding mechanisms of routers. Conse quently, any changes in addressing schemes such as address space extension do not require changes in the AOPS nodes. It can also support both connectionoriented and connec tionless services to carry various types of traffic such as ATM and IP traffic. This structure is mainly based on the hierar chical source routing approach. The analytical results show that average packet header sizes are still acceptable even for long paths consisting of many nodes each of which has a large number of output ports.
The future is photons: no more electrons; we know most core networks will eventually have to be photonic. We know most access (whether in voice, or TV or Internet) is moving to wireless. Much light is handy to make nets work, even if photons can be quite big and unwieldy.
In that context, this paper revisits a combination of some oldish ideas. Essentially, addresses are constructed in such a way as to be a source route, for a “self-routing” network – each hop has a lambdaslot interchange table, which the next part of the address indexes directly, thus allowing very simple logic to process each packet. If the address space is large enough, and IPv6 addresses appear to be good enough, then fairly arbitrary networks can be source/self routed this way.
The idea is pretty much like a mix of concepts from ATM and Manhatten Grid routing nets, as well as wormhole routed nets, but extended to the wide area. Forwarding table index field based routing was part of Paul Francis’ next generation Internet proposal (PIP) for IPv6, and was documented in a paper and his thesis.
The paper is timely because various groups are looking at how the photonic core rolls out in the Internet in a way to permit maximum flexibility. The paper is well written and presented, and contains useful background calculations (e.g. packet header size) to show that the schemes proposed fit within both topological (diameter/hop count of future optical cores) and technological constraints (optical processing) that we might expect, and shows how the Inter-domain case can be handled within the same framework.