Joint Reactive and Proactive SDN Controller Assignment for Load Balancing

Gouareb, Racha; Friderikos, V.; Aghvami, A.H.; Tatipamula, Mallik

doi:10.1109/gcwkshps45667.2019.9024555

Cited by 3 publications

(4 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An ant colony with an external memory CPP algorithm was proposed in [14] and validated against particle swarm optimization (PSO) for load balancing. In [32], the authors explored the load-balancing CPP with a predetermined number of controllers, addressing proactive and reactive controller-node assignments. In [15], a two-stage framework was developed to optimize energy consumption and task allocation using PSO and deep reinforcement learning.…”

Section: Literature Reviewmentioning

confidence: 99%

“…The CPP research commonly conceptualized the system as an undirected graph G(V, E), where the set of candidate locations for controllers surrounded controlled nodes V. These locations were modeled as a subset of V connected to controlled nodes via wired links represented as the set of edges E in the network graph [4]- [10], [13], [16], [24]- [29], [31], [32]. Alternatively, some works adopted set theory for system modeling [14], [15], defining pairs of device-todevice wireless links, wireless network entities, and SDN controllers using various sets.…”

Section: Literature Reviewmentioning

confidence: 99%

“…Overall, the CPP, categorized as an NP-hard optimization problem, has been approached with various approximation algorithms and heuristic methods across different network architectures, ranging from wired to wireless. The literature underscores two main categories of optimal placement methods: heuristic approaches [4], [5], [7], [9], [10], [14], [15], [17], [25]- [28], [30], [31] and deterministic approaches integrating approximation algorithms [6], [8], [13], [16], [24], [29], [31], [32].…”

Section: Literature Reviewmentioning

confidence: 99%

See 2 more Smart Citations

On Robust Optimal Joint Deployment and Assignment of RAN Intelligent Controllers in O-RANs

Abdel-Rahman,

Mazied,

Hassan

et al. 2024

IEEE Open J. Commun. Soc.

View full text Add to dashboard Cite

The open radio access network (O-RAN) architecture is consolidating the concept of softwaredefined cellular networks beyond 5G networks, mainly through the introduction of the near-real-time radio access network (RAN) intelligent controller (Near-RT RIC) and the xApps. The deployment of the Near-RT RICs and the assignment of RAN nodes to the deployed RICs play a crucial role in optimizing the performance of O-RANs. In this paper, we develop a robust optimization framework for joint RIC deployment and assignment, considering the uncertainty in user locations. Specifically, our contributions are as follows. First, we develop C 3 P 2 , a robust static joint RIC placement and RAN node-RIC assignment scheme. The objective of C 3 P 2 is to minimize the number of RICs needed to control all RAN nodes while ensuring that the response time to each RAN node will not exceed δ milliseconds with a probability greater than β. Second, we develop CPPA, a robust joint RIC placement and adaptive RAN node-RIC assignment scheme. In contrast to C 3 P 2 , CPPA enjoys a recourse capability, where the RAN node-RIC assignment adapts to the variations in the user locations. We use chance-constrained stochastic optimization combined with several linearization techniques to develop a mixed-integer linear (MIL) formulation for C 3 P 2 . Two-stage stochastic optimization with recourse, combined with several linearization techniques, is used to develop an MIL formulation for CPPA. The optimal performance of C 3 P 2 and CPPA has been examined under various system parameter values. Furthermore, sample average approximation has been employed to design efficient approximate algorithms for solving C 3 P 2 and CPPA. Our results demonstrate the robustness of the proposed stochastic resource allocation schemes for O-RANs compared to existing deterministic allocation schemes. They also show the merits of adapting the allocation of resources to the network uncertainties compared to statically allocating them.INDEX TERMS Open radio access networks (O-RANs), RAN intelligent controller (RIC) placement, RIC assignment, chance-constrained stochastic optimization, two-stage stochastic optimization with recourse, sample average approximation.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

On Robust Optimal Joint Deployment and Assignment of RAN Intelligent Controllers in O-RANs

Abdel-Rahman,

Mazied,

Hassan

et al. 2024

IEEE Open J. Commun. Soc.

View full text Add to dashboard Cite

show abstract

“…The reactive load balancing algorithm generates a new optimal path after the loss, whereas the proactive load balancing algorithm monitors the network status periodically and computes optimal link. Thus, the proactive approach tries to avoid data loss, 8 and it saves the waiting time of the incoming flow from the path computation time. Hence, we designed a proactive algorithm to avoid the data loss and improve the network performance.…”

Section: Introductionmentioning

confidence: 99%

Intelligence‐enabled approach for load balancing in software‐defined data center networks

Fancy

Pushpalatha

2021

Int J Communication

View full text Add to dashboard Cite

Summary Software‐defined network provides eminent solutions for many complex network management functionalities in a data center network (DCN). One of the major tasks in any network is the load balancing in the available links. Due to dynamic data traffic nature in network, it is necessary to perform deep learning approaches for the long‐term and the short‐term data. This paper proposes a splitting policy‐based RL network (SPRLN) approach, a reinforcement learning‐based proactive load balancing algorithm that avoids the poll to the controller after the switch encounters an abnormality. The proposed method has been tested with simulations and found successful in improving the overall network performance by taking appropriate action for reward maximization. The testbed environment is treated as a Q‐learning algorithm; here, the optimality is defined as the path having the least score so that the overloaded path in that particular time can be avoided. An artificial neural network is needed because the data are uncertain all the time. Thus, the proposed SPRLN method yields 30% increased throughput and with 80% reduced data loss, when compared to existing approaches.

show abstract