Distributed hash tables (DHTs) are a class of decentralized distributed systems that can efficiently search for objects desired by the user. However, a lot of communication traffic comes from multi-word searches. A lot of work has been done to reduce this traffic by using bloom filters, which are space-efficient probabilistic data structures. There are two kinds of bloom filters: fixed-size and variable-size bloom filters. We cannot use variablesize bloom filters because doing so would mean wasting time to calculating hash values. On the other hand, when using fixed-size bloom filters, all the nodes in a DHT are unable to adjust their false positive rate parameters. Therefore, the reduction of traffic is limited because the best false positive rate differs from one node to another. Moreover, in related works, the authors took only two-word searches into consideration, In this paper, we present a method for determining the best false positive rate for three- or more word searches. We also used a new filter called a ringed filter, in which each node can set the approximately best false positive rate. Experiments showed that the ringed filter was able to greatly reduce the traffic.