Research Article
Markov Approximation for Task Offloading and Computation Scaling in Mobile Edge Computing
Algorithm 1
The Markov approximation-based algorithm.
| (1) | The following procedures execute on the MD. | | (2) | procedure Initialization | | (3) | for do | | (4) | , and , | | (5) | Randomly assign m to the MD (i.e., ) or a MEC server (i.e., ) | | (6) | end for | | (7) | Randomly pick a feasible from | | (8) | Transit to STAGE 1 | | (9) | end procedure | | (10) | procedure Stage 1: Wait (m) | | (11) | Generate an exponentially distributed timer with mean equal to γ | | (12) | Begin counting down by | | (13) | end procedure | | (14) | procedure Stage 2: Hop (m) | | (15) | If expires then | | (16) | Send RESET signals to , . | | (17) | Generate a new configuration by randomly selecting another feasible device for m | | (18) | if the target device is the MD then | | (19) | , and , | | (20) | Randomly pick another feasible from | | (21) | else if the target device is a MEC server then | | (22) | , , and , | | (23) | end if | | (24) | Migrate to the new configuration by the probability of | | (25) | end if | | (26) | end procedure | | (27) | procedure Stage 3: Reset (m) | | (28) | if does not expire and m receives a RESET message then | | (29) | Terminate its current countdown timer | | (30) | Transit to STAGE 1 | | (31) | end if | | (32) | end procedure |
|
