2 Establishing Task Migration Model
There are two ways to perform tasks on the mobile end. One is to process directly at the
mobile end, and the other is to put tasks on MEC server for processing and return the
results to the local area. If the local processing time of a task is less than or equal to the
time when it is transferred to the remote, then it is processed locally. Otherwise, the
task is sent to remote processing. The decision is expressed by n k;s :
n k;s ¼
0; task k don
0 t translate by channels
1; task k computing in cloud by channels
ð2:1Þ
Suppose that the set of user devices is DN ¼ 1; 2; 3; . . .; DN
j j
f
g , the set of transmission
channels
is
CN ¼ 1; 2; 3; . . .; CN
j j
f
g ,
the
priority
set
is
PN ¼ 1; 2; 3; . . .; PN
j j
f
g . For a task k, CW k represents the workload (CPU cycles), DS k
represents the data size of the task, OD k represents the output size after processing and
PN k represents the priority of the task. Task k can be represented by quadruples:
W k ¼ CW k ; DS k ; OD k ; PN k
f
g .
2.1 Task Processing Time Locally
Let f
l
k represents the rate at which task k executes on the local device (CPU cycles/per
sec), so the time required for task k to execute locally can be expressed as:
T
l
k ¼
CW k
f l
k
ð2:2Þ
2.2 Task Processing Time in MEC Server
2.2.1 Transmission Delay of Tasks in Remote Processing
When a task is processed on the MEC server, the data transmission rate of task
k transmitted through the channel s can be expressed as [3]:
R k;s n
ð Þ ¼ w s log 2 1 þ
P
tr
k G k
r 2 þ
P
h2DN i
f g:n h;s ¼1;h6 ¼k P
tr
h Á G h
!
ð2:3Þ
where r
2 is the transmission noise and G k is the channel gain of the k-th device. In this
case, we assume that in an open environment, G k ¼ l
Àa , where l is the distance between
the device and the base station, a is the path fading factor, and l
Àa is the fading power
with distance. Therefore, the transmission delay of task k through channel s is:
t
tr
k;s ¼
DS k
R k;s
ð2:4Þ
Research on Multi-priority Task Scheduling Algorithms …
141
There are two ways to perform tasks on the mobile end. One is to process directly at the
mobile end, and the other is to put tasks on MEC server for processing and return the
results to the local area. If the local processing time of a task is less than or equal to the
time when it is transferred to the remote, then it is processed locally. Otherwise, the
task is sent to remote processing. The decision is expressed by n k;s :
n k;s ¼
0; task k don
0 t translate by channels
1; task k computing in cloud by channels
ð2:1Þ
Suppose that the set of user devices is DN ¼ 1; 2; 3; . . .; DN
j j
f
g , the set of transmission
channels
is
CN ¼ 1; 2; 3; . . .; CN
j j
f
g ,
the
priority
set
is
PN ¼ 1; 2; 3; . . .; PN
j j
f
g . For a task k, CW k represents the workload (CPU cycles), DS k
represents the data size of the task, OD k represents the output size after processing and
PN k represents the priority of the task. Task k can be represented by quadruples:
W k ¼ CW k ; DS k ; OD k ; PN k
f
g .
2.1 Task Processing Time Locally
Let f
l
k represents the rate at which task k executes on the local device (CPU cycles/per
sec), so the time required for task k to execute locally can be expressed as:
T
l
k ¼
CW k
f l
k
ð2:2Þ
2.2 Task Processing Time in MEC Server
2.2.1 Transmission Delay of Tasks in Remote Processing
When a task is processed on the MEC server, the data transmission rate of task
k transmitted through the channel s can be expressed as [3]:
R k;s n
ð Þ ¼ w s log 2 1 þ
P
tr
k G k
r 2 þ
P
h2DN i
f g:n h;s ¼1;h6 ¼k P
tr
h Á G h
!
ð2:3Þ
where r
2 is the transmission noise and G k is the channel gain of the k-th device. In this
case, we assume that in an open environment, G k ¼ l
Àa , where l is the distance between
the device and the base station, a is the path fading factor, and l
Àa is the fading power
with distance. Therefore, the transmission delay of task k through channel s is:
t
tr
k;s ¼
DS k
R k;s
ð2:4Þ
Research on Multi-priority Task Scheduling Algorithms …
141