Protocol Selection Sample Clauses

Protocol Selection. The specific nature of group communication in a Bluetooth network is an important factor that co-determines the protocol choice. Bluetooth networks (called Piconets) use three-bit active device adressing and thus are limited to 8 active devices in a group. Piconet devices do not have direct links be- tween each other; instead, they share a two-way link with the device that is designated as a master in that network, forming a star topology. The master’s clock determines the communication intervals for both the slaves (devices connected to the master) and itself. While a slave may com- municate with the master without restrictions, slave-to-slave communica- tion is facilitated by the master, who has to resend the message to the correct recipient. In terms of message complexity, this means that any message whose recipient is not the master himself is sent twice – once from the member to the leader, and then from the leader to the correct recipient. Slave-to-slave broadcasts thus have to be transmitted by the master as n 2 messages, each sent to one of the slave devices in the network. This effectively means that even constant-round protocols will finish in a linear number of rounds in a real-world implementation. It is therefore desirable to minimize the number of broadcasts with respect to the overall number of messages exchanged during the protocol run. Another important factor is the overall computational complexity of the protocol. While it is desirable to keep the amount of computation as low as possible, this should not come at the expense of an inordinate increase of communication complexity. A good example of this are the protocols from the ▇▇▇▇▇▇▇▇▇-▇▇▇▇▇▇▇ family, such as ▇▇▇▇▇-▇▇▇▇▇ or ▇▇▇▇-▇▇▇▇. Due to the high number of simultaneous broadcasts in each of the turns, a Blue- tooth implementation would require the master to send O(n2) number of messages in every turn. With regards to the selection criteria outlined above, we chose to imple- ment a modified version of the Augot et al. protocol, and the unmodified version of the GDH.3 protocol. In addition to good computational com- plexity, the Augot protocol maps very well to the underlying master-slave topology as members (slaves) only communicate with the leader (master). This amounts to a total of 3n 3 messages exchanged during the protocol run. GDH.3, on the other hand, is slightly less computationally demanding than the Augot protocol at the expense of more messages exchanged during the protoco...