MPR redundancy

The core functionality section of RFC3626 states that MPRs should be chosen so that all 2 hop nodes are covered by at least one MPR. This selection scheme will result in highly optimized flooding. But once again, bandwidth optimization can be sacrificed for robustness. One could decide that 2 hop neighbors should be covered by more than one MPR if possible. To do this, a parameter MPR_COVERAGE is introduced. This parameter specifies how many MPRs the MPR calculation should attempt to set for a 2 hop neighbor.

Core-MPR calculation states that every two hop neighbor must be covered by at least one MPR. One can not transform this rule to be every two hop neighbor must be covered by at least n MPRs while working with MPR redundancy. This is due to the fact that two hop neighbors might not be reachable through more than one symmetric neighbor. MPR redundancy becomes the attempt to get two hop neighbors covered by up to n MPRs.

Figures 4.5 and 4.6 illustrates the usage of the MPR_COVERAGE parameter. As seen in figure 4.6 incrementing MPR_COVERAGE leads to less optimized retransmission.

Figure: The center node has chosen the black nodes as MPRs using a MPR_COVERAGE parameter of 1.

Figure: The center node has chosen the black nodes as MPRs using a MPR_COVERAGE parameter of 2. Not all two hop neighbors are covered by two MPRs since they are only reachable through one one hop neighbor.