2

As a hobby, I am writing a simple and primitive distributed web search engine. It occurred to me that it currently has no protection against malicious peers trying to skew the search results.

The principle idea of the project is:

  • To store the inverse indices and ranking factors in a Kad (Kademlia) DHT (Distributed Hash Table).
  • The peers update the inverse indices as they crawl web.

I used Google Scholar to find a solution. It seems that most of the authors of proposed P2P web search ignores the above-mentioned problem.

I think I need some kind of reputation system or trust metrics. My knowledge in this domain is sufficiently lacking. I would very much appreciate a few pointers.

AJed
  • 2,432
  • 19
  • 25
Moonwalker
  • 155
  • 3

1 Answers1

1

When tackling such problems, think in general terms. Researchers have studied the impact of malicious peers in numerous papers. Most of these systems should work.

However, define your problem clearly. Ex: What are the malicious peers trying to do in this case? do you want to:

  1. protect your system from unwanted outsiders (security),
  2. protect your data from faulty peers (fault tolerance),
  3. protect your data from greedy peers (greedy peers), or others?

Also note that Kademlia is very similar to Chord, SkipNet, Pastry and others (some researchers call these networks "common prefix networks - I can't remember", others: ring-based p2p networks, etc ... Some solutions may be designed specifically for these networks. However, I don't see why a solution made for a structured P2P networks would not work for your architecture.

AJed
  • 2,432
  • 19
  • 25