Software Defined Approaches for Management of Mobility, Load and Energy in Wireless Networks

Rangisetti, A K (2017) Software Defined Approaches for Management of Mobility, Load and Energy in Wireless Networks. PhD thesis, Indian Insitute of Technology Hyderabad.

Text
CS12P1001.pdf - Submitted Version
Restricted to Registered users only until 28 June 2020.
Download (1MB) | Request a copy

Abstract

Proliferation of smart mobile devices and their applications demand Mobile Network Operators (MNOs) to expand their infrastructure to address coverage and capacity issues. Cellular networks, with their current in fl exible and expensive network infrastructure, are facing various challenges in ef fi ciently handling the exponentially growing traf fi c demands of users. MNOs have started dense deployment of Long Term Evolution (LTE) Heteroge- neous Networks (HetNets) with various small cells (Femto Base Stations (FBSs), pico, mi- cro, etc. ) under overlaying Macro cells for expanding network coverage and offering higher data rates. To further augment capacity of cellular networks, Wi-Fi Access Points (APs) are also being deployed by MNOs. However, dense deployment of small cells increases con- trol plane complexity in handling Radio Access Network (RAN) tasks like load balancing, interference management, mobility management and energy savings. In distributed LTE- RANs, in order to solve RAN control plane tasks ef fi ciently, various cells need to exchange a lot of messages over X2 interface. Moreover, deployment of a lot of small cells can lead to higher energy consumption. Besides, changing existing solutions or incorporating new solutions can lead to increase in capital expenditure (CAPEX) and operational expenditure (OPEX) of operators. Aforementioned issues raised need of simpli fi cation of control and management tasks and ef fi cient usage of radio resources in wireless networks such as LTE HetNets and IEEE 802.11 based Wireless Local Area Networks (WLANs). This could be achieved by adopting novel networking paradigms which could simplify the task of net- work management and control tasks, and allow faster deployment of newer solutions on top of existing network hardware equipment by software updates/upgrades. Software De- fi ned Networking (SDN) is one such revolutionary paradigm which makes networks more agile and fl exible by separation of data plane and control plane tasks. In this thesis work, in order to offer seamless mobility in multi-channel enterprise WLANs environment, a programmable WLAN architecture is used. Speci fi cally, a seam- less load-aware hand-off algorithm is proposed and its performance is evaluated on a Soft- ware De fi ned WLAN (SD-WLAN) testbed. Proposed load-aware hand-off algorithm not only offers seamless mobility in enterprise WLAN, it also effectively utilizes neighbor APs for offering load balance in the network. Besides, in order to provide programmable, fl exi- ble and scalable solutions for LTE-RAN, a Software De fi ned-LTE-Radio Access Network (SD-LTE-RAN) framework is proposed using OpenFlow enabled eNodeBs (OFeNBs). Proposed SD-LTE-RAN framework is implemented in NS-3 simulator with OpenFlow module and then used for evaluating the performance of various load balance algorithms and cell switch-off mechanisms. Exponential growth of mobile traf fi c can cause exorbitant load even in LTE networks. vii As User Equipments (UEs) are typically associated with a near-by cell (eNB), spatio- temporal variation in traf fi c demands makes the LTE networks suffer from load imbalance problem. Due to the distributed nature of eNB operation in LTE-RAN, traditional solu- tions to tackle load imbalance problem could lead to excessive overhead over X2 interface. Hence, managing densely deployed cells is very challenging in the existing distributed LTE RAN. In this thesis work, load imbalance issue is addressed by proposing a novel Quality of Service (QoS) Aware Load Balance (QALB) algorithm on the SD-LTE-RAN framework. For taking load balance decisions, the QALB algorithm considers loads of neighbor cells, QoS pro fi les of UEs and their expected throughputs w.r.t. neighbor cells. Unlike existing load balance algorithms, it does not change handover-offset parameters of cells to avoid ping pong handovers. The QALB runs in linear time in terms of number of cells and UEs, and hence it is suitable for real-time deployment of LTE networks. In various load bal- ance experiments conducted in NS-3, proposed QALB algorithm is able to maintain better QoS data rates ( > 80% of their con fi gured Guaranteed Bit Rates (GBRs)) for more than 70% of the cells in the network, while existing load balance solutions are able to do the same for only 50% of the cells in the network. In overall, the QALB algorithm is able to decrease the total network overload by 15% compared to existing solutions. We also eval- uated the QALB algorithm in mobility scenarios and identi fi ed that it is able to decrease average network overload by 10% compared to existing solutions. To evaluate the network wide fair load distribution, we de fi ned Load Balance Index (LBI) using Jain’s Fairness In- dex and found that QALB is also able maintain better LBI compared to existing solutions. Moreover, control ovehead of QALB and existing load balance algorithms is same on the proposed framework. In this thesis work, in order to reduce energy costs of the HetNets, we propose an in- terference and QoS aware cell switch-off strategy (IQ-CSOS) on the SD-LTE-RAN frame- work. The IQ-CSOS runs in polynomial time in terms of number of Macros, small cells and UEs. Unlike existing CSOSs, in selection of cells for switch-off and subsequent han- dover of their associated UEs to overlaying Macros, IQ-CSOS considers both traf fi c load of small cells and their cross-tier interference effect on the Macros. Hence, IQ-CSOS is able to provide higher energy savings. In evaluation, unlike existing works, we investi- gated both network energy costs and QoS satisfaction of sessions during CSOS decisions. In performance evaluation carried out in NS-3 under various test scenarios, it is observed that IQ-CSOS is able to provide 50-80% of network energy savings in terms of small cells energy consumption. Besides, it is able to provide 30% more energy savings compared to existing CSOSs with marginal affect on QoS in the network. Moreover, control ovehead of IQ-CSOS and existing CSOSs is same on the proposed framework

[error in script]

IITH Creators: