Advances, Systems and Applications
Table 18 Comparison of load-balancing techniques in SD-IoT
From: An overview of QoS-aware load balancing techniques in SDN-based IoT networks
Disadvantages | Advantages | Layer | Function | Technique |
|---|---|---|---|---|
- Communication overhead - Security challenges |  + Support user mobility  + Minimizing the number of active servers  + Reduce the possibility of rejecting tasks due to resource constraints  + Energy management and response time reduction  + Improving resource utilization | Cloud / Fog / Control | Services movement between resources | Migration |
- Challenges in finding a reliable route in mobile networks - Routing overhead |  + Traffic management  + Improvement of some QoS parameters | Cloud / Fog | Finding the best path to transfer tasks and their data | Routing |
- Rerouting overhead |  + Low overhead and failure probability by choosing the right path  + Prevent network congestion  + Improving reliability, energy, delay, throughput, packet delivery rate, and resource utilization | Cloud / Fog | Finding alternative routes | Rerouting |
- Architectural design costs |  + Improving the quality of service  + Workload management  + Reducing complexity and increasing efficiency of control units  + Achieving distributed control of flows for scalability and reliability | Cloud / Fog / Control | Changing the centralized control layer to a distributed one | Architecture |
- Possibility of overloading and underloading resources |  + Policies related to cache, congestion control, queuing, scheduling, green computing, and security  + Improving the QoS | Cloud / Fog / Control | Change in network configuration and management | Policy |
- Probability of violating resource capacity threshold - Lack of scalability |  + Improving response time, throughput, and delay  + Network congestion control  + Support of mobile applications  + Maximizing request acceptance rate and optimizing the use of resources | Fog | Delivery of tasks to other resources for network balance and stability | Offloading |
- Challenges in determining the number of clusters - Local minimum problem - Computational cost |  + Increasing stability  + Effective resource management  + Minimum communication cost between tasks  + Improving load-balancing, scalability, availability, integrity, and security | Cloud / Fog / Control | Group tasks and resources based on their similarities | Clustering |
- Lack of trade-off between speed and accuracy of the traffic classification mechanism |  + Possibility of reliable communication  + Network traffic forecasting  + Improving load-balancing, response time, throughput, and resource utilization  + Providing QoS in routers  + Separation of traffic in different streams  + Filtering and intrusion detection  + Allocation of appropriate levels of QoS to different applications | Cloud / Infrastructure | Classification of traffic based on the type of tasks and their requirements based on predefined rules | Classification |
- Possibility of resource overhead |  + Balance between performance and QoS  + Creating justice on network nodes  + Maximize network capacity | Cloud / Fog / Control | Mapping tasks on available resources Based on resource status | Allocation |
- Increase in rejection of tasks |  + Network congestion control  + Improving delay, packet delivery ratio, and packet loss ratio | Fog | Responsible for ensuring the authentic network load | Admission control |
- Increased delay |  + Reducing network traffic load  + Extending the lifetime of the network in energy-constraint networks  + Reducing the number of packets sent in the network  + Reducing communication costs and accurate data recovery  + Reduce data redundancy | Cloud / Edge | Collecting and combining data from different sources | Aggregation |
- Temporal mismatch between traffic load and resource availability - Balance and orchestration of virtual resources |  + Increase utilization and reduce costs  + Efficient resource management  + Increase security  + Reducing energy consumption | Cloud / Fog | Implementing virtual functions on physical resources | Virtualization |
- Taking too much time on large-scale controller placement - Link setup latency for a switch to controller communications - Possible delay in placing controllers |  + Development of receptive capacity  + Increased network fault tolerability  + Management flexibility  + Improve overall network performance  + Minimize runtime and delay  + Choosing the best number of controllers | Control / Edge | Deployment of the controller in a suitable place | Controller placement |
- Reduced security - Inefficiency in saturation scenarios |  + Congestion control  + Reduced response time  + Ensuring a proportional share of traffic for better use of resources | Fog | Splitting traffic into multiple paths | Flow change |
- User categorization based on the application requirement - Prediction of user behaviour and energy requirements |  + Reducing energy costs for end users  + Reducing peak energy consumption  + Reducing the congestion of transmission lines | Infrastructure | Users' energy consumption management in response to resource conditions | User demand management |
- Need to analyze the temporal-spatial - Mobility management in heterogeneous networks |  + Dynamic resource management  + Improving task completion time, throughput, and delay | Fog | Efficient and appropriate assignment of resources based on task execution time | Scheduling |