A sparse sampling algorithm for self-optimisation of coverage in LTE networks

Thampi, Ajay; Kaleshi, Dritan; Randall, Peter N.; Featherstone, W.; Armour, Simon

doi:10.1109/iswcs.2012.6328500

Cited by 12 publications

(9 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Sparse sampling is another technique that can handle the-curse-of-dimensionality problem [17]. This method was utilized for the self-optimization of the coverage through the antenna tilt optimization [18]. The recent advances in deep RL reveal that neither fuzzy RL nor sparse sampling is as efficient as deep learning (DL) based models.…”

Section: A Related Work and Motivationmentioning

confidence: 99%

Online Antenna Tuning in Heterogeneous Cellular Networks With Deep Reinforcement Learning

Balevi

Andrews

2019

IEEE Trans. Cogn. Commun. Netw.

View full text Add to dashboard Cite

We aim to jointly optimize antenna tilt angle, and vertical and horizontal half-power beamwidths of the macrocells in a heterogeneous cellular network (HetNet). The interactions between the cells, most notably due to their coupled interference render this optimization prohibitively complex. Utilizing a single agent reinforcement learning (RL) algorithm for this optimization becomes quite suboptimum despite its scalability, whereas multi-agent RL algorithms yield better solutions at the expense of scalability. Hence, we propose a compromise algorithm between these two. Specifically, a multi-agent mean field RL algorithm is first utilized in the offline phase so as to transfer information as features for the second (online) phase single agent RL algorithm, which employs a deep neural network to learn users locations. This two-step approach is a practical solution for real deployments, which should automatically adapt to environmental changes in the network. Our results illustrate that the proposed algorithm approaches the performance of the multi-agent RL, which requires millions of trials, with hundreds of online trials, assuming relatively low environmental dynamics, and performs much better than a single agent RL. Furthermore, the proposed algorithm is compact and implementable, and empirically appears to provide a performance guarantee regardless of the amount of environmental dynamics.

show abstract

Section: A Related Work and Motivationmentioning

confidence: 99%

Online Antenna Tuning in Heterogeneous Cellular Networks With Deep Reinforcement Learning

Balevi

Andrews

2019

IEEE Trans. Cogn. Commun. Netw.

View full text Add to dashboard Cite

show abstract

“…incapability of handling a large set of network configurations and the inability to adapt to the network environment without prior knowledge, a centralized RL sparse sampling algorithm is proposed in [85]. The authors of [85] focus on the problem of self-optimization in the LTE network environment by adjusting antenna tilt and show that this approach is more efficient than supervised learning and Q-learning algorithms in terms of self-healing performance and multiple coverage problems. The authors in [86] add dynamic and adaptive antenna tilt adjustment for the best trade-off between coverage and capacity in mobile networks.…”

Section: Antenna Tilt Approach (Directional Antenna)mentioning

confidence: 99%

Interference Management Techniques in Small Cells Overlaid Heterogeneous Cellular Networks

Samal

2018

JMM

View full text Add to dashboard Cite

Now-a-days because of the rapid increase in mobile users mobile data has also raised. Due to the heavy demand of data rates for advanced applications service providers are forced to adopt different technological advancements in conventional networks. Looking toward this context heterogeneous networks (HetNets) play a vital role in future 5G wireless cellular network deployment. In Heterogeneous Cellular Networks (HCNs) low power small cells are overlaid with the existing macro-only cell which increases the complexity of the network. Some of the emerging network technologies for this cellular evolution are the Femtocell, Picocell and Metrocell networks. Due to the deployment of these small cells in HCNs there are wide varieties of network structures and different transmitting powers which lead to interference in the cellular network. In addition to this, due to the spatiotemporal distribution of the mobile users and their mobility, hotspots also lead to increase of the possibilities of interference generation in wireless cellular network systems. The major technical issues associated with the mass deployment of smaller cells are related to the interference management between smallcells i.e. Femtocells and other serving cells in the same spectrum. Since Femtocells and Macrocells share the same frequency bands, it is necessary to develop efficient interference management techniques to increase the capacity and throughput of HCNs. So in order to meet the user demands in an efficient way

show abstract

“…In order to cope with NP-Hard nature of the problem, these works have mainly resorted to heuristics such as tabu-search [8], fuzzy reinforcement learning [15], fuzzy q-learning [23], golden section search [28], Taguchi method [20], multi-level random Taguchi's method [24], reinforcement learning based sparse sampling [25] and simulated annealing [26]. The general methodology followed in these works has been to evaluate the desired Key Performance Indicator(s) (KPIs) as a function of system-wide tilt angles through a simulation model.…”

mentioning

confidence: 99%

Self Organization of Tilts in Relay Enhanced Networks: A Distributed Solution

Imran

Abu-Dayya

et al. 2014

IEEE Trans. Wireless Commun.

View full text Add to dashboard Cite

Despite years of physical-layer research, capacity enhancement potential of relays is limited by the additional spectrum required for Base Stations (BS)-Relay Station (RS) links. This paper presents a novel distributed solution by exploiting a system level perspective instead. Building on a realistic system model with impromptu RS deployments, we develop an analytical framework for tilt optimization that can dynamically maximize spectral efficiency of both BS-RS and BS-user links in online manner. To obtain a distributed self-organizing solution, the large scale system-wide optimization problem is decomposed into local small scale subproblems by applying the design principles of self-organization in biological systems. The local subproblems are non-convex but having a very small scale can be easily solved via standard nonlinear optimization techniques such as sequential quadratic programming. The performance of developed solution is evaluated through extensive simulations for LTE-A type system and compared against number of benchmarks including a centralized solution obtained via brute force, that also gives an upper bound to assess optimality gap. Results show that proposed solution can enhance average spectral efficiency by up to 50% compared to fixed tilting with negligible signaling overheads. The key advantage of the proposed solution is its potential for autonomous and distributed implementation.

show abstract

A sparse sampling algorithm for self-optimisation of coverage in LTE networks

Cited by 12 publications

References 2 publications

Online Antenna Tuning in Heterogeneous Cellular Networks With Deep Reinforcement Learning

Online Antenna Tuning in Heterogeneous Cellular Networks With Deep Reinforcement Learning

Interference Management Techniques in Small Cells Overlaid Heterogeneous Cellular Networks

Self Organization of Tilts in Relay Enhanced Networks: A Distributed Solution

Contact Info

Product

Resources

About