| 1. |
- Arvidsson, Ake, et al.
(författare)
-
Fast Transport for Edge Computing in 5G Networks
- 2018
-
Ingår i: 2018 26TH INTERNATIONAL CONFERENCE ON SOFTWARE, TELECOMMUNICATIONS AND COMPUTER NETWORKS (SOFTCOM). - : IEEE. ; , s. 41-45
-
Konferensbidrag (refereegranskat)abstract
- We consider TCP performance for cloud services in 5G cellular networks and show that edge clouds offer the highest throughput, but that further improvements still are possible. To this end we devise a concept where edge clouds connected directly to the radio access network can increase their transmission rates by relying on the built in re-transmissions (through quality of service features) and on the built in user fairness (through per-user buffers and scheduler policies). We use numerical examples from measurements in real networks to demonstrate that the concept is efficient and significantly improves the effective throughput beyond what placement alone can do, and we also outline possible further improvements.
|
|
| 2. |
- Arvidsson, A., et al.
(författare)
-
Transport bottlenecks of edge computing in 5G networks
- 2019
-
Ingår i: Journal of Communication Software and Systems. - : University of Split. - 1845-6421 .- 1846-6079. ; 15:1, s. 59-65
-
Tidskriftsartikel (refereegranskat)abstract
- We consider bottlenecks of TCP throughput in scenarios with edge clouds and 5G cellular networks. By numerical examples from measurements in real networks, we show that edge clouds indeed improve throughput but that several, non-negligible bottlenecks remain. We therefore devise a concept where edge clouds connected directly to the radio access network can increase their transmission rates by relying on built in retransmissions (through quality of service features) and on the built in user fairness (through per-user buffers and scheduler policies). We then return to the numerical examples and show that our solution provides substantial gains and we conclude by identifying and discussing the remaining bottlenecks and the potential of an improved protocol.
|
|
| 3. |
|
|
