We can use different heavy hitter thresholds $T_i$ based on some pre-agreed strings agg_param ∈ agg_parameters by the servers.
Example from PLASMA paper: This can be beneficial for traffic avoidance since different roads may have different traffic densities. For example, highways are busier than smaller suburban roads. The servers can take that into consideration during evaluation, and use higher $T$ s for highways (since there are more vehicles), and lower thresholds for smaller roads.
The aggregation parameters can be tuples instead of single values:
meaning that the threshold $T$ for aggregation_param = 0b0000 is 7, the $T$ for 0b0001 is 6, and so on. Nice to have: if an aggregation parameter does not have an explicit threshold (like (0b0000000, _)) it should inherit from its parent (so it'd be 7 in this case).
We can use different heavy hitter thresholds $T_i$ based on some pre-agreed strings
agg_param ∈ agg_parametersby the servers.Example from PLASMA paper: This can be beneficial for traffic avoidance since different roads may have different traffic densities. For example, highways are busier than smaller suburban roads. The servers can take that into consideration during evaluation, and use higher $T$ s for highways (since there are more vehicles), and lower thresholds for smaller roads.
The aggregation parameters can be tuples instead of single values:
meaning that the threshold $T$ for
aggregation_param = 0b0000is 7, the $T$ for0b0001is 6, and so on. Nice to have: if an aggregation parameter does not have an explicit threshold (like(0b0000000, _)) it should inherit from its parent (so it'd be 7 in this case).