Service Provisioning Update Scheme for MobileApplication Users in a Cloudlet Network

Service Provisioning Update Scheme for Mobile
Application Users in a Cloudlet Network
Huawei Huang #, Song Guo*
# The University of Aizu, Japan
*Department of Computing, The Hong Kong Polytechnic University
Email: davyhwang.cug@gmail.com, song.guo@polyu.edu.hk

Topic and Scenario Background
• This paper studies the problem of adaptive service
provisioning strategy towards mobile users in the
context of big data application.
• The workload generated by these applications is
supposed power-hungry, and computing-intensive.
• Therefore, the workload needs to be offloaded to
other entities such as
• a cloudlet network,
• or a remote private cloud.

Our Motivations
• (User’s characteristics) When mobile users are
using a (big data) application with their devices,
e.g., , they are
• getting online or offline at different time
• moving in different areas of a metropolitan at different
time
• An example of service provisioning for mobile
users under a cloudlet based network is shown in
Fig. 1.
• How to achieve the cost-minimized workload-
offloading at different time-frame for each user,
becomes an interesting problem.
Fig. 1 Dynamic characteristics of an edge network, e.g., a mobile
user alternates status between online and offline frequently.

Major contributions
• In this paper,
• we study an adaptive service provisioning problem under the cloudlet based
network, where we try to find a near optimal update scheme to decide how
to update the provisioning solution for each mobile user at each time-frame if
the trajectory of each mobile user is provided.
• With the objective to maximize a weighted profit for network operators, we
first formulate this problem to a nonlinear programming problem, which is
then transformed to a solvable integer linear programming using the absolute
value manipulation techniques.
• A framework of algorithms is proposed.

System model
• In a hybrid network with:
• cloudlet network, and
• remote private cloud;
• a set of users U, and
• a set of cloudlet servers S
• Variables to denote:
Indicator of Inter-cloud VM-migration event:
Where to deploy a VM for user u:

Formulation
• Mathematic model
Letting denote the access delay of user u in U at
time-slot t in T:
we can compute the total access delay at time-slot t
in T:
Letting ζ denotes the normalized VM-migration delay
between the private cloud and a local cloudlet server, Then,
the total VM-migration cost of all users at ts t in T:
Then, a profit-maximization is formulated as
the following nonlinear programming:
The objective is to maximize a weighted
profit, which positively associates with the
overall admitted traffic rate that is served by
the local cloudlet network and negatively
associates with the total access delay and the
migration delay.
Cons. (4b) expresses that the capacity of each server
should not be expired.
(4c) describes the relationship between variables X and Z.
(4d) imposes the detailed special rule for var zu
t when user
u is in each time-slot of set F(u).

Formulation (cont.)
• Mathematic model
Cons. (4c) involves the absolute value functions,
making this problem nonlinear and unsolvable
using linear programming methods.
Cons. (4b) expresses that the capacity of each server
should not be expired.
(4c) describes the relationship between variables X and Z.
(4d) imposes the detailed special rule for var zu
t when user
u is in each time-slot of set F(u).
Therefore, we particularly transform (4c) to two linear
constraints through the following manipulation of
absolute value expression:
Finally, the original profit-maximization is re-formulated to:

The proposed framework of heuristic
algorithms
• To find solutions X and Z, the basic idea of the
algorithm, to construct a solution, subject to a
feasibility specification.
• The structure of a solution and the feasibility
specification:

NUMERICAL EVALUATION
• Parameter Settings:
• The network adopted in our simulations is a cloudlet based urban access network
with 10 adjacent macrocells.
• We randomly generate a traffic demand trace for each mobile user within [10,
100] Mb/s.
• access delay to the remote cloud is fixed to 10 ms
• local access delay of any mobile user to its located cloudlet server is randomly
generated within [1, 3] ms
• the inter-cloud VM-migration delay is normalized to 10 ms.
• We then generate a sequential trajectory for each mobile user with 20 time slots.
• In each time-slot, the probability of any mobile user stays in online status is
set to 0.8.

Results
• Metrics: Profit, Cloudlet traffic rate,
Weighted access delay, Weighted
migration delay
• Algorithms: First-Fit-Decreasing, First-Fit-
Increasing, Online-First-Decreasing and
Online-First-Increasing, which are
represented by FFD, FFI, OFD and OFI,
respectively.
• We also solve the formulation to retrieve
the Optimal solution using Gurobi 6.0
[23], under each simulation setting.
Fig. 4. Performance of algorithms when varying the serving capacity
of local server (i.e., Cs) in a range 600-1500 Mb/s.
In a summary, we can always observe that the FFI and
OFI have a similar performance and outperform other
two heuristics in terms of all the metrics.

Service Provisioning Update Scheme for MobileApplication Users in a Cloudlet Network

Recommended

More Related Content

What's hot (19)

Similar to Service Provisioning Update Scheme for MobileApplication Users in a Cloudlet Network (20)

Recently uploaded (20)