FemtoClouds (CC, Resource Management)

Question 1

Background (Femtoclouds)?

Answer 1

• Femtoclouds are dynamic, decentralised edge computing networks. They are clusters of co-located IoT/Mobile devices able to handle the computation of tasks both within and outside of the cluster.
• The concept builds upon earlier work in edge computing, including cyberforaging, Cloudlets, and Fog computing.
• Femtoclouds leverage the resources of underutilized mobile edge devices.

Question 2

Problem (FemtoClouds)?

Answer 2

• Traditional client-server deployments have substantial inertia.
• Cloud services have increased latency and bandwidth costs compared to LAN environments.
• Existing edge computing setups are expensive, requiring the purchase of cloudlet servers.
• Mobile devices have limited resources that go unutilized.

Question 3

Solution (FemtoClouds)?

Answer 3

• Femtoclouds efficiently distribute and execute computational tasks across “helper devices”.
• A cloud-based controller manages compute resources and interfaces with job originators.
• Job managers handle individual jobs, looking ahead of the computation stage and allocating helpers for each task.
• Femtoclouds use a hybrid edge-cloud architecture.

Question 4

Applications/Uses (FemtoClouds)?

Answer 4

None

Question 5

Strengths and Weaknesses (FemtoClouds)?

Answer 5

Strengths:
- Lower latency, resulting in faster response times.

• Privacy and security achieved by keeping data more local.
• Improved scalability due to the dynamic, decentralised architecture.
• Reduced network congestion as task execution is closer to the data source.
• Cheaper than deploying dedicated Cloudlet-servers.
• Leverage unused compute power.
• Adaptive to system changes due to central controller modelling.

Weaknesses:
Transferring large amounts of data to and from the cloud is not fast or reliable.