[ndnSIM] Question about How to Modify Packet Format and Producer Processing for the Packet

大熊 浩也 okuma at stu.kobe-u.ac.jp
Mon Sep 2 20:26:36 PDT 2019



On 2019/09/03 10:45, Lixia Zhang wrote:
> 
> 
>> On Sep 2, 2019, at 9:41 AM, Hiroya Okuma <okuma at stu.kobe-u.ac.jp> wrote:
>>
>> Thank you for your reply.
>>
>> First, the network model consists of servers (providers), routers, and consumers. Servers and routers are wired, and routers and consumers are connected by Wi-Fi.
>>
>> Each node also caches regular files (A, B, C, etc.) and encoded files (A + B, A + B + C, etc.).
> 
> 1/ Let us be precise here: each node caches *data packets* (not files); a data packet may be carrying a segment of a file.


I was misrecognized. Thank you for teaching me.
Does it feel like file A is divided into (A1, A2, ..., An) and each is 
stored in a data packet?
How does this affect my network model in that each node caches * data 
packet * instead of a file?

> 
> 2/ for a data packets that carries a piece of network-encoded data: how is that data packet named?

Since simple simulations are assumed, all files to be prepared have the 
same size.
The data packet of encoded file A + B XORed with file A and file B is 
going to be named ("A1 + B1", "A2 + B2", ..., "An + Bn") .

> 
>> Therefore, if a consumer with files B and C requests file A, file A can be decrypted using its own cache file even if A + B or A + C is obtained.
>> Therefore, the consumer sends an Interest packet that contains three files name [A, A + B, A + C].
>> In the scenario, the node (router or server) that received the Interest packet returns only one to the consumer if it has any of the three requested files.
>>
>> If you have any other questions, please.
>>
>> 2019/09/02 23:32、Lixia Zhang <lixia at cs.ucla.edu <mailto:lixia at cs.ucla.edu>>のメール:
>>
>>>
>>>> On Sep 1, 2019, at 11:35 PM, 大熊 浩也 <okuma at stu.kobe-u.ac.jp <mailto:okuma at stu.kobe-u.ac.jp>> wrote:
>>>>
>>>> Thank you for answering.
>>>> I will ask you some more questions.
>>>>
>>>> On 2019/09/02 14:59, Xinyu Ma wrote:
>>>>> Hello, Ookuma san,
>>>>>> On Sep 1, 2019, at 10:48 PM, 大熊 浩也 <okuma at stu.kobe-u.ac.jp <mailto:okuma at stu.kobe-u.ac.jp>> wrote:
>>>>>>
>>>>>> Hello everyone.
>>>>>>
>>>>>> I have some problems that I have to solve quickly.
>>>>>> Could you give me some advice?
>>>>>> (Let me apologize for my poor English and Technical capabilities)
>>>>>>
>>>>>> ・How to Modify Interest Packet Format and Producer Processing for the Packet
>>>>>> I am studying the combination of ICN and coding network.
>>>>>> I'm thinking consumers send a single Interest packet that contains a request file name and file names that can be decrypted by using the contents of consumers own content store.
>>>>>> (Since the producer will only return one of the requested files, it may be difficult for the producer to process if the consumer sends multiple Interest packets?)
>>>>>> So I want to store multiple request packet names in Interest packets.
>>>>>>
>>>>> I’m sorry I don’t quite understand this. If you want multiple files, why not just send out multiple packets?
>>>>> I don’t think merging Interests is a good idea because it will make the Content Store useless.
>>>>
>>>> Not requesting multiple files, I want to store several request file candidates in the Interest packet.
>>>> Example) When a consumer who owns two files B and C and requests file A
>>>> Interest packet ← "A", "A + B", "A + C" (Request file A can be decrypted if one of these files is obtained)
>>>>
>>>> When sending multiple Interest files, if the producer returns all the requested files, will the consumer receive more files than necessary, causing congestion?
>>>
>>> it seems that we need to step up a level to get a better understand of the scenario, before dive into the question of what kind of names one can put into an Interest packet.
>>>
>>> So the first question to clarify is what kind of network coding is being tried here?
>>>
>>> Lixia
>>>
> 
> 

-- 
********************************************
大熊 浩也 (Hiroya Okuma)
神戸大学 工学研究科 電気電子工学専攻
情報通信研究室所属
E-mail: okuma at stu.kobe-u.ac.jp
********************************************




More information about the ndnSIM mailing list