Rational Server Selection for Mobile Agents

In mobile edge computing, a set of edge servers is geographically deployed near the mobile users such that accessible computing capacities and services can be provided to users with low latency. Due to user’s mobility, one fundamental and critical problem in mobile edge computing is how to select edge servers for many mobile users so that the total waiting time is minimized. In this paper, we propose a multi-user waiting time computation model about composite services and show the resource contention of the edge server among mobile users. Then, we introduce a novel and optimal Multi-user Edge server Selection method based on Particle swarm optimization (MESP) in mobile edge computing, which selects edge servers for mobile uses in advance within polynomial time. Extensive simulations on a real-world data-trace show that the MESP algorithm can effectively reduce the total waiting time compared with traditional approaches.

Download Full-text

Joint routing and server selection for multiple description video streaming in ad hoc networks

IEEE International Conference on Communications, 2005. ICC 2005. 2005 ◽

10.1109/icc.2005.1494940 ◽

2005 ◽

Cited By ~ 2

Author(s):

Shiwen Mao ◽

Xiaolin Cheng ◽

Y.T. Hou ◽

H.D. Sherali ◽

J.H. Reed

Keyword(s):

Ad Hoc Networks ◽

Video Streaming ◽

Ad Hoc ◽

Server Selection ◽

Multiple Description ◽

Selection For ◽

Hoc Networks

Download Full-text

Multiple channel queues in heavy traffic. III: random server selection

Advances in Applied Probability ◽

10.2307/1426325 ◽

1970 ◽

Vol 2 (2) ◽

pp. 370-375 ◽

Cited By ~ 22

Author(s):

Ward Whitt

Keyword(s):

Queueing System ◽

Heavy Traffic ◽

Multiple Channel ◽

Server Selection ◽

Selection For ◽

Selection Probabilities ◽

Service Channels

As in [4] and [5], we study service facilities with r arrival channels and s service channels. However, here we assume that customers, immediately upon arrival, randomly select one of the s service channels. Successive customers make this choice independently, choosing server i with probability pi, p1 + · · · + ps = 1. Customers are then served by the servers they select in order of their arrival without defections. The average processing rates as well as the server selection probabilities may vary from server to server, but again we assume the r arrival channels are independent and independent of the service channels. The service channels are not independent, however, because of the random server selection. For simplicity, we only consider a single queueing system; the extension to sequences follows immediately using the argument of [5].

Download Full-text