D2K_magration_strategy () |
{ |
Functional analysis of our D2K strategy can be performed through a triplet format, |
that is, , where I is the input to the functions, P is the process |
to describe the internal operations, and O is the output of the functions. |
Data_ accumulation () |
{ |
Input: raw data at motion (flow) |
Output: raw data at rest (storage) |
Process: use pull/push method to accumulate data |
If (method = push) |
{Sensors periodically sense environment and send data to CH} |
If (method = pull) |
{CH injects SQL into sensor environment to extract data} |
CH performs data validations to check the correctness of the data |
} |
Replica_ elimination () |
{ |
Input: raw sensor dataset |
Output: distinct dataset |
Process: data matching |
If (old data value = new data value) |
{CH discards to store the new value and the old value remains in the CH storage} |
If (old data value != new data value) |
{CH stores the new data value into its storage} |
} |
Data_ calibration () |
{ |
Input: distinct dataset |
Output: valid range dataset |
Valid data range = [lb, ub] |
// ub defines the upper threshold value |
// lb defines the lower threshold value |
Process: equivalence class partitioning |
If (ub < distinct data < lb) |
{CH treats these data as an invalid data class} |
If (lb <= distinct data <= ub) |
{CH treats these data as a valid data class} |
CH rejects the invalid data class from its storage |
} |
Data_ fusion () |
{ |
Input: valid dataset |
Output: integrated database |
Process: database operation |
While (arrival of valid data = true) |
{ |
CH maps the data into the database with its access key |
} |
} |
Knowledge_ filtration () |
{ |
Input: integrated database |
Output: control instructions to be stored in the knowledge base |
Process: fuzzy controller operation with a FIS |
While (availability of integrated database = true) |
{CH permits the database into a fuzzy controller circuitry embedded |
with a FIS to extract the useful knowledge} |
} |
} |