Radio access network sharing
View Patent ↗A method implemented in an apparatus used in a radio access network (RAN) sharing system including a plurality of basestations is disclosed. The method includes estimating resource requirement or demand of one or more entities in each base station according to feedback from the plurality of basestations, computing resource allocation for said one or more entities, and enforcing the computed resource allocation using basestation-level virtualization. Other methods, apparatuses, and systems also are disclosed.
1. A method implemented in an apparatus used in a radio access network (RAN) sharing system including a plurality of basestations, the method comprising:
estimating resource requirement or demand of one or more entities in each base station according to feedback from the plurality of basestations;
computing resource allocation for said one or more entities; and
enforcing the computed resource allocation using basestation-level virtualization,
wherein the estimation comprises:
computing the resource requirement or demand for entity j at base station b using the following formula:
d
j
b
=
∑
i
∈
Q
j
b
min
(
β
A
i
,
S
i
)
T
×
R
i
,
where d b j is the resource requirement or demand, Q b j is a set of flows that belong to entity j at basestation b, A i and S i are the arrival rate and maximum sustained rate, respectively, of flow i that belongs to entity j, β is a parameter, T is the total number of resource blocks in a base station in one second, and R i is the average number of bits per resource block derived from per-flow MCS information.
2. The method as in claim 1 , wherein the resource allocation is computed dynamically or the method is performed periodically.
3. The method as in claim 1 , wherein the feedback comprises per-flow MCS (modulation and coding scheme) information.
4. The method as in claim 1 , wherein the computation comprises:
formulating a problem as a constrained convex optimization problem;
finding a basically feasible solution which violates an individual upper bound;
finding a feasible solution using a phase-one method; and
finding an optimal solution using a barrier method.
5. The method as in claim 4 , wherein the formulation comprises:
using objective function g j (t) which represents the total utility each entity obtains when receiving aggregate t resources,
wherein each entity has lower bound L j and upper bound U j on the aggregate resources,
each basestation has total normalized resource less than or equal to 1, and
resource allocated to entity j (j=1, 2, . . . , J) at basestation b (b=1, 2, . . . , B) has individual upper bound u b j and lower bound l b j .
6. The method as in claim 5 ,
wherein individual lower bound l b j =αL j /B, where 0<α<1, and
wherein individual upper bound u b j is chosen to be proportional to the estimated resource requirement or demand.
7. The method as in claim 4 , wherein the formulation comprises:
maximizing
∑
b
=
1
B
∑
j
=
1
J
G
j
,
b
(
t
j
b
)
such that
L
j
≤
∑
b
=
1
B
t
j
b
≤
U
j
for
all
j
,
∑
j
=
1
J
t
j
b
≤
f
r
(
b
)
for
all
b
,
and
l
j
b
≤
t
j
b
≤
u
j
b
,
for
all
b
,
j
,
where t b j is resource allocation for entity j (j=1, 2, . . . , J) at base station b (b=1, 2, . . . , B), G j,b (t b j ) is a utility function, L j and U j are lower and upper bounds on the aggregate resources, respectively, l b j and u b j are individual lower and upper bounds, respectively, and fr(b) represents normalized resources available at basestation b.
8. The method as in claim 7 , wherein G j,b (t b j ) is expressed as follows:
G j,b ( t b j )= d b j ×log( t b j ).
9. The method as in claim 7 , wherein G j,b (t b j ) is expressed as follows:
G j,b ( t b j )= d b j ×t b j .
10. The method as in claim 4 , wherein the basically feasible solution is obtained by using the following formula:
t
j
b
=
l
j
b
+
(
δ
j
+
η
j
)
·
1
-
∑
j
l
j
b
B
-
∑
j
,
b
l
j
b
,
where t b j is resource allocation for entity j (j=1, 2, . . . , J) at base station b (b=1, 2, . . . , B),
δ
j
=
L
j
-
∑
b
l
j
b
,
and
0
≤
η
j
≤
min
(
(
B
-
∑
j
L
j
)
/
J
,
U
j
-
L
j
)
.
11. The method as in claim 4 , wherein the phase-one method comprises:
starting with the basically feasible solution;
replacing individual upper bound u b j with u b j +s b j , where s b j ≧0; and
minimizing
∑
b
,
j
s
j
b
.
12. The method as in claim 1 , wherein the apparatus is in a cellular gateway or between the cellular gateway and one of the plurality of basestations.
13. An apparatus used in a radio access network (RAN) sharing system including a plurality of basestations, the apparatus comprising:
an estimation unit to estimate resource requirement or demand of one or more entities in each base station according to feedback from the plurality of basestations;
a computing unit to compute resource allocation for said one or more entities; and
an enforcing unit to enforce the computed resource allocation using basestation-level virtualization,
wherein the resource requirement or demand is computed for entity j at base station b using the following formula:
d
j
b
=
∑
i
∈
Q
j
b
min
(
β
A
i
,
S
i
)
T
×
R
i
,
where d b j is the resource requirement or demand, Q b j is a set of flows that belong to entity j at basestation b, A i and S i are the arrival rate and maximum sustained rate, respectively, of flow i that belongs to entity j, β is a parameter, T is the total number of resource blocks in a base station in one second, and R i is the average number of bits per resource block derived from per-flow MCS information.
14. The apparatus as in claim 13 , wherein the feedback comprises per-flow MCS (modulation and coding scheme) information.
15. The apparatus as in claim 13 ,
wherein, for the computation, the apparatus formulates a problem as a constrained convex optimization problem, finds a basically feasible solution which violates an individual upper bound, finds a feasible solution using a phase-one method, and finds an optimal solution using a barrier method.
16. The apparatus as in claim 13 , wherein the apparatus is in a cellular gateway or between the cellular gateway and one of the plurality of basestations.
17. A method used in a radio access network (RAN) sharing system, the method comprising:
transmitting feedback from a plurality of basestations; and
computing resource allocation for one or more entities in each of the plurality of basestations according to the feedback,
wherein the feedback includes per-flow MCS (modulation and coding scheme) information,
wherein the computation comprises:
formulating a problem as a constrained convex optimization problem;
finding a basically feasible solution which violates an individual upper bound;
finding a feasible solution using a phase-one method; and
finding an optimal solution using a barrier method, and
wherein the formulation comprises:
using objective function g j (t) which represents the total utility each entity obtains when receiving aggregate t resources,
wherein each entity has lower bound L j and upper bound U j on the aggregate resources,
each basestation has total normalized resource less than or equal to 1, and
resource allocated to entity j (j=1, 2, . . . , J) at basestation b (b=1, 2, . . . , B) has individual upper bound u b j and lower bound l b j .
18. The method as in claim 17 , wherein the resource allocation is computed dynamically or periodically.
19. The method as in claim 17 , wherein the apparatus is in a cellular gateway or between the cellular gateway and one of the plurality of basestations.