Mobile Edge Computing (MEC) and Network Function Virtualization (NFV) are emerging as key technologies to meet the demands of an expanding number of users in 5G. In NFV, the placement of Virtual Network Functions (VNFs) is a crucial decision as it significantly impacts the performance, reliability, and cost of the network. VNF placement and optimization play a vital role in ensuring the efficient and effective operation of NFV-enabled MEC nodes.
Service Function Chaining (SFC) is a network architecture that facilitates the sequential processing of network traffic through a set of service functions. SFC involves forwarding network traffic to a predefined order of service functions, such as firewalls and load balancers, identified by policy. This process provides a specific set of services to the traffic. SFC routes packets through a chain of network functions before reaching their destination.
The primary motivation for NFV-enabled MEC networks is to address the limitations of traditional hardware-based network appliances. It enhances network agility, allowing for the easy introduction of new network services. Service function chains provide a flexible and efficient framework for deploying network functions in a 5G network. SFC enables network operators to deploy network functions in a modular manner, offering a flexible framework for deploying functions in a dynamic and changing network environment. SFC provides a way to automate the setup of virtual network connections to handle different types of traffic flows.
To design and implement cost-effective placement and chaining of Virtual Network Functions (VNFs) in Multi-access Edge Computing (MEC) environment to create a Service Function Chain (SFC) such that latency constraints of every SFC are met.
- To place VNFs in MEC servers such that resource constraint of every VNF is met.
- To create Service Function Chains for the optimally placed VNFs in an edge computing environment as per the requirements of various applications.
- To create an optimal Service Function Chain and satisfy the latency constraints of every SFC.
| MEC | CPU | RAM | Processor Rate | Number of Resources |
|---|---|---|---|---|
| MEC1 | 48 | 88 | 7 | 12 |
| MEC2 | 32 | 86 | 9 | 8 |
| MEC3 | 30 | 80 | 6 | 9 |
| MEC4 | 50 | 72 | 10 | 5 |
| MEC5 | 36 | 84 | 7 | 5 |
| VNF | CPU | RAM |
|---|---|---|
| VNF1 | 2 | 8 |
| VNF2 | 4 | 16 |
| VNF3 | 8 | 32 |
| VNF4 | 6 | 48 |
| VNF5 | 10 | 20 |
| VNF1 | 5 | 25 |
| VNF2 | 4 | 20 |
| VNF3 | 8 | 50 |
| VNF4 | 2 | 35 |
| VNF5 | 8 | 36 |
| VNF1 | 10 | 48 |
| VNF2 | 6 | 20 |
| VNF3 | 8 | 40 |
| VNF4 | 6 | 40 |
| VNF5 | 8 | 48 |
| MEC | Allocations |
|---|---|
| MEC1 | [VNF2, VNF4, VNF5] |
| MEC2 | [VNF2, VNF3, VNF2] |
| MEC3 | [VNF1, VNF5, VNF1] |
| MEC4 | [VNF4, VNF5, VNF1] |
| MEC5 | [VNF3, VNF4, VNF3] |
Query: [VNF2, VNF5, VNF4, VNF3, VNF1, VNF3]
| Algorithm | Reward Function Value | Time(in sec) | Service Function Chain |
|---|---|---|---|
| Q-learning | 34.56262 | 0.310210 | [MEC2, MEC1, MEC4, MEC2, MEC4, MEC5] |
| SARSA | 40.05592 | 0.289654 | [MEC2, MEC1, MEC1, MEC2, MEC3, MEC2] |
VNF Placement in MEC using Genetic algorithms improves the performance of edge computing networks by maximizing the deployment of virtual network functions near end users. Low-latency service function chaining leads to a faster response time and reduced delay, supporting many applications such as online gaming, video streaming, etc. We have implemented VNF placement using the Genetic algorithm and Service Function Chaining using Q-learning and SARSA reinforcement algorithms. We have combined the results of VNF placement with the SARSA algorithm and obtained optimal Service Function Chaining, ensuring low latency. The proposed SARSA model is compared with the standard Q-Learning model. The comparison concludes that the proposed model outperforms the already existing Q-Learning model for VNF placement and SFC chaining in the Multi-access Edge Computing environment.