US 11,961,409 B1 · Air-ground joint trajectory planning and offloading scheduling method and system for distributed multiple objectives

Abstract

An air-ground joint trajectory planning and offloading scheduling method and system for distributed multiple objectives is provided. At the beginning of each timeslot, an unmanned aerial vehicle (UAV) selects a flight direction based on a total energy consumption of all devices and a total amount of unprocessed data of all the devices in the current system, and flies a fixed distance towards a certain direction. Before the UAV reaches a new location, each terrestrial user independently selects a task data offloading scheduling strategy based on the total energy consumption of all the devices and the total amount of the unprocessed data of all the devices in the current system. In order to improve an expected long-term average energy efficiency and data processing capability, the present disclosure also provides average feedbacks for an energy consumption and unprocessed data.

An air-ground joint trajectory planning and offloading scheduling method for distributed multiple objectives, wherein in the method, an unmanned aerial vehicle (UAV) fixed at a flight altitude and a terrestrial base station (BS) at a fixed location jointly serve a plurality of terrestrial users at fixed locations, and the method specifically comprises the following steps:

step 1:
obtaining, by the UAV, current state information, wherein the state information comprises a location of the UAV in a current timeslot t and a total amount of unprocessed data of all devices in a previous timeslot t−1, and all the devices comprise the UAV, the BS, and the plurality of users;
step 2:
selecting, by the UAV, an action, selecting a flight direction from a preset flight direction set by minimizing a total energy consumption of all the devices and a total amount of unprocessed data of all the devices, and flying a fixed distance;
wherein the total energy consumption of all the devices is calculated according to a following formula:
D
t
=
∑
m
=
1
m
=
M
(
D
m
,
t
)
+
D
UAV
,
t
+
D
B
⁢
S
,
t
wherein Dt represents a total energy consumption of all the devices in the timeslot t, M represents a quantity of users, m∈{1, 2, . . . , M}, and Dm,t, DUAV,t, and DBS,t respectively represent cumulative computing tasks in queues of a user m, the UAV, and the BS in the timeslot t;
and
the total amount of the unprocessed data of all the devices is calculated according to a following formula:
E
t
=
∑
m
=
1
M
(
(
E
UAV
,
m
,
t
trans
+
E
UAV
,
m
,
t
cp
)
·
α
m
,
t
,
UAV
+
(
E
BS
,
m
,
t
trans
+
E
BS
,
m
,
t
cp
)
·
α
m
,
t
,
BS
+
E
UE
,
m
,
t
cp
·
α
m
,
t
,
UE
)
wherein Et represents a total amount of unprocessed data of all the devices in the timeslot t, EUAV,m,ttrans and EBS,m,ttrans respectively represent transmission energy consumptions of offloading a task by the user m to the UAV and the BS in the timeslot t, EUAV,m,tcp and EBS,m,tcp respectively represent energy consumptions of computing the task by the user m on the UAV and the BS in the timeslot t, and αm,t,UE, αm,t,UAV, and αm,t,BS respectively represent operations of performing local computing, offloading to the UAV, and offloading to the BS by the user m in the timeslot t, wherein when the local computing is performed, αm,t,UE=1, otherwise αm,t,UE=0;
when the offloading to the UAV is performed, αm,t,UAV=1, otherwise αm,t,UAV=0;
and when the offloading to the BS is performed, αm,t,BS=1, otherwise αm,t,BS=0;
step 3:
before the UAV reaches a new location, obtaining, by each user, current state information;
step 4:
selecting, by all the users, an action in parallel, determining a task offloading scheduling method of each user from a preset offloading strategy set by minimizing the total energy consumption of all the devices and the total amount of the unprocessed data of all the devices, and executing the task offloading scheduling method in the current timeslot t;
step 5:
after task offloading scheduling of the users is completed in the current timeslot t, updating the location of the UAV as a location in a next timeslot t+1, wherein the UAV and each user receive a feedback for a current action;
wherein the feedback for the current action is as follows:
r
_
=
[
E
¯
,
D
¯
]
T
⁢
E
¯
=
lim
T
→
∞
sup
⁢
1
T
⁢
ε
⁢
{
∑
t
=
0
T
-
1
e
t
}
⁢
D
¯
=
lim
T
→
∞
sup
⁢
1
T
⁢
ε
⁢
{
∑
t
=
0
T
-
1
d
t
}
wherein r represents a column vector of the feedback, [ ]T represents transposition, Ē represents an average feedback for an energy consumption, D represents an average feedback for unprocessed data, T represents duration of the task,
sup
⁢
1
T
represents an upper bound of
1
T
,
, ε{ } represents averaging, et=−Et, and dt=−Dt;
and
step 6:
repeating steps 1 to 5 in each timeslot to obtain an optimal flying and offloading strategy based on the state information and the feedback for the current action, wherein the flying and offloading strategy comprises the flight direction of the UAV and the task offloading scheduling method of each user.

193 more elements

Air-ground joint trajectory planning and offloading scheduling method and system for distributed multiple objectives

Life of this patent

Air-ground joint trajectory planning and offloading scheduling method and system for distributed multiple objectives

Life of this patent

Abstract

Claims6