hello every one,
I am wondering if ndn does not use any information of users for sending Interest message, how this network deliver data to the user.Moreover, I would like to say that the Data message is passed the reverse path that interest message is received and each router saves the interface from where it receives this interest message. any how, if more than one user is connected to one interface ( for example in wireless network), the router broadcasts data message to all of them? if this can be happened, how they can control behavior of user and block the malicious user ?
any help will be appreciated
thanks in advance
best regards
In the wireless environment, the PIT can be enhanced to include previous hop instead of incoming interface like in the wired environment. In this way, the Data packet can be sent back to the right node.
Seems to me that users are identified by a name, as are the sources of information. The user's name is needed to get the information back, on the reverse path.
The presumed benefit of this is that you cannot exhaust the "address space." But to me, this is not sensible. The name is an address, and one way or another, the size of each name will have to be limited, ultimately.
The name has to consist of characters. The characters are a combination of binary bits (8 bits per character? 16 bits per character? Whatever, right?). One way or another, you will be defining addresses consisting of some number of bits, although in principle that number of bits may be variable, and positively delimited by some means. Whatever the case, practical limits to name lengths will exist, I have to believe.
If the routers on the forward path store the information from the source of the interest message, this mean that all routers in the path must maintain state information. This is called "PIT state" in NDN terminology. But it's not a good idea for scalability, and certainly a very bad idea in case a router in the path fails before the information is returned. Stateless routing was one significant innovation introduced by IP. I wouldn't think it's a liability. On the other hand, if the names are unique and well known, then you're back to what we have now with IP, with some other address size limitation.
As to how information is found, when the user creates a name, I don't think that's much different from what search engines do today. Clever algorithms. Still a topic of research.
Strange concept. In some ways, the existence of search engines, combined with the IP Domain Name System (DNS), have functionally created from IP networks what these content-centric networks want to achieve. Consider, for example, that a totally unformatted "interest" request a user sends to a search engine generates a series of alternative "names." And then the "names," also known as URLs, of variable length but delimited with a set of conventions, are fed to the DNS. Multiple names sent to the DNS may actually all point to one address, in some cases, much as in these NDNs. And the DNS then returns the address or addresses (that the user never actually sees).
So maybe the idea is to do away with the DNS, and instead create "addresses" out of names. Always good to try to relate the new concept to what we already know. (Plus ça change, plus c'est la même chose.)
thanks for giving your idea . any how ,Truong-Xuan Do, if we consider previous hop , we need to consider identity for each user, otherwise, how we can save the previous hop. moreover, the considering identity like IP address can not be a good idea , since it will degrade the benefit of ndn than IP network.
- Badges
- Science topic