AMAndrei Marinescu

Andrei Marinescu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Andrei Marinescu has a total of 23 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Learning-Based Remote Channel Inference: Feasibility Analysis and Case Study

    by Sheng Chen, Zhiyuan Jiang, Sheng Zhou, Zhisheng Niu, Ziyan He, Andrei Marinescu, Luiz A. DaSilva

    Channel state information (CSI) plays a vital role in wireless communication systems. However, the CSI acquisition overhead is an enormous obstacle to realize the system performance improvements promised by massive connectivity and massive multiple-input-multiple-output (MIMO). To alleviate this overhead, this paper proposes a remote channel inference framework by probing the channels occupied by a source base station (BS) and inferring the channels of target BSs at geographically separated sites. The work generalizes existing literature which mainly focuses on utilizing the CSI linear correlations of adjacent antennas, by adopting a model-free deep learning framework to investigate non-linear dependence among remote CSI. The existence of such cross-BS CSI dependence is first shown by calculating the mutual information between remote channels, and the Cramér-Rao lower bound of remote CSI inference performance based on a one-ring channel model. Inspired by this finding, modern deep learning approaches are leveraged to perform remote channel inference in heterogeneous networks for both single user and multi-user scenarios. The simulation results based on ray tracing data show evident performance advantages over conventional methods, under both homogeneous and heterogeneous frequency coverage. The proposed framework achieves beamformer inference accuracy within 4.6% of the genie-aided optimum at the cost of sweeping only two beams.


  • A Multi-Agent Neural Network for Dynamic Frequency Reuse in LTE Networks

    by Andrei Marinescu, Irene Macaluso, Luiz A. DaSilva

    Fractional Frequency Reuse techniques can be employed to address interference in mobile networks, improving throughput for edge users. There is a tradeoff between the coverage and overall throughput achievable, as interference avoidance techniques lead to a loss in a cell's overall throughput, with spectrum efficiency decreasing with the fencing off of orthogonal resources. In this paper we propose MANN, a dynamic multiagent frequency reuse scheme, where individual agents in charge of cells control their configurations based on input from neural networks. The agents' decisions are partially influenced by a coordinator agent, which attempts to maximise a global metric of the network (e.g., cell-edge performance). Each agent uses a neural network to estimate the best action (i.e., cell configuration) for its current environment setup, and attempts to maximise in turn a local metric, subject to the constraint imposed by the coordinator agent. Results show that our solution provides improved performance for edge users, increasing the throughput of the bottom 5% of users by 22%, while retaining 95% of a network's overall throughput from the full frequency reuse case. Furthermore, we show how our method improves on static fractional frequency reuse schemes.


  • Open Access

    A Multi-Agent Neural Network for Dynamic Frequency Reuse in LTE Networks

    by Andrei Marinescu, Irene Macaluso, Luiz A. DaSilva

    Fractional Frequency Reuse techniques can be employed to address interference in mobile networks, improving throughput for edge users. There is a tradeoff between the coverage and overall throughput achievable, as interference avoidance techniques lead to a loss in a cell's overall throughput, with spectrum efficiency decreasing with the fencing off of orthogonal resources. In this paper we propose MANN, a dynamic multiagent frequency reuse scheme, where individual agents in charge of cells control their configurations based on input from neural networks. The agents' decisions are partially influenced by a coordinator agent, which attempts to maximise a global metric of the network (e.g., cell-edge performance). Each agent uses a neural network to estimate the best action (i.e., cell configuration) for its current environment setup, and attempts to maximise in turn a local metric, subject to the constraint imposed by the coordinator agent. Results show that our solution provides improved performance for edge users, increasing the throughput of the bottom 5% of users by 22%, while retaining 95% of a network's overall throughput from the full frequency reuse case. Furthermore, we show how our method improves on static fractional frequency reuse schemes.

  • Residential demand response: Experimental evaluation and comparison of self-organizing techniques

    by Ivana Dusparic, Adam Taylor, Andrei Marinescu, Fatemeh Golpayegani, Siobhan Clarke


  • Open Access

    System Level Evaluation and Validation of the ns-3 LTE Module in 3GPP Reference Scenarios

    by Andrei Marinescu, Irene Macaluso, Luiz A. DaSilva

    LTE networks represent the latest generation implementation of communications technologies for mobile users. As such, simulation tools are essential in the development of new radio resource management (RRM) techniques for LTE. In this paper, we evaluate the LTE module of the ns-3 simulator in a set of 3GPP reference scenarios for system level simulation. Downlink spectral efficiency, user throughput distribution and SINR distribution in a basic LTE configuration are compared against 3GPP results, which are based on an aggregate of 17 industrial simulators. Results show that the ns-3 LTE module achieves similar performance to the one obtained by the 3GPP industrial simulators in the evaluated cases, both in terms of SINR distributions and users' throughput.


  • Open Access

    Prediction-Based Multi-Agent Reinforcement Learning in Inherently Non-Stationary Environments

    by Andrei Marinescu, Ivana Dusparic, Siobhán Clarke

    Multi-agent reinforcement learning (MARL) is a widely researched technique for decentralised control in complex large-scale autonomous systems. Such systems often operate in environments that are continuously evolving and where agents’ actions are non-deterministic, so called inherently non-stationary environments. When there are inconsistent results for agents acting on such an environment, learning and adapting is challenging. In this article, we propose P-MARL, an approach that integrates prediction and pattern change detection abilities into MARL and thus minimises the effect of non-stationarity in the environment. The environment is modelled as a time-series, with future estimates provided using prediction techniques. Learning is based on the predicted environment behaviour, with agents employing this knowledge to improve their performance in realtime. We illustrate P-MARL’s performance in a real-world smart grid scenario, where the environment is heavily influenced by non-stationary power demand patterns from residential consumers. We evaluate P-MARL in three different situations, where agents’ action decisions are independent, simultaneous, and sequential. Results show that all methods outperform traditional MARL, with sequential P-MARL achieving best results.


  • Maximizing renewable energy use with decentralized residential demand response

    by Ivana Dusparic, Adam Taylor, Andrei Marinescu, Vinny Cahill, Siobhan Clarke

    Due to steady urbanization, the electrical grid is facing significant changes in the supply of resources as well as changes in the type, scale, and patterns of residential user demand. To ensure sustainability and reliability of electricity provision in the growing cities, a significant increase in energy generated from renewable sources (e.g., wind, solar) is required. However, renewable energy supply is much more variable and intermittent than traditional supply, as it depends on changing weather conditions. In order to optimize residential energy usage, demand response (DR) techniques are being investigated to shift device usage to the periods of low demand. Currently most DR approaches focus on traditional DR goals, e.g., reducing usage at peak times and increasing it at off-peak times. More flexible and adaptive techniques are needed that can not only meet traditional DR requirements, but enable just-in-time use of renewable energy, rather than requiring its curtailment or using expensive and inefficient storage options. This paper proposes the use of decentralized learning-based multi-agent residential DR to enable more efficient integration of renewable energy sources in the smart grid, in the presence of increased demand caused by high electric vehicle penetration. We evaluate the approach using real household usage data obtained from Irish smart meter trials and data on wind-generated energy from the Irish grid operator. We discuss advantages of the proposed decentralized approach and show that it is able to respond to multiple variable wind-generation patterns by shifting up to 35% of the overall energy usage to the periods of high wind availability.


  • Open Access

    Self-organising algorithms for residential demand response

    by Adam Taylor, Ivana Dusparic, Colin Harris, Andrei Marinescu, Edgar Galvan-Lopez, Fatemeh Golpayegani, Siobhan Clarke, Vinny Cahill

    Residential Demand Response has shown promising results in smart grid applications. It can be achieved manually or autonomously. The variety of algorithms applied to achieve autonomous Demand Response have lacked a common baseline. The selection of testing environment is typically skewed by the desire to find one in which a particular algorithm performs well. This work evaluates several algorithms in a common evaluation environment. Which has been designed to encompass the range of conditions in a typical Residential Demand Response application. The environments described exhibit the following characteristics: non-stationary, dynamic, multi-actor, multi-objective. These characteristics will then be used to provide heuristics for algorithm selection. The algorithms used were selected to cover the spectrum of possible approaches to Demand Response. Some are centralised, others distributed. There are collaborative approaches and non-collaborative ones. Some are learning based, others require no training. This work provides criteria for which particular algorithms should be applied to a given application.


  • A dynamic forecasting method for small scale residential electrical demand

    by Andrei Marinescu, Ivana Dusparic, Colin Harris, Vinny Cahill, Siobhan Clarke

    Small scale electrical demand forecasting is an emerging field motivated by the penetration of renewable energy sources and the growth of microgrids and virtual power plants. These advances pose more complex forecasting challenges compared to the already established large scale forecasting approaches. Current short term load forecasting methods deal with two types of day, normal and anomalous, which are predicted separately. Anomalous days are classified as such ahead of time, based on key calendar events such as public holidays. However, there are some anomalous days which are not always predictable on a day ahead basis. Due to unforeseen events, a seemingly normal day can progress towards an anomalous case causing high errors in prediction. We propose a new dynamic forecasting mechanism that actively monitors residential electrical demand along a forecasted day, and detects anomalous pattern changes from a previously predicted demand of the day. A self-organising map is employed to detect anomalous days as they progress. Once an anomaly is detected, a neural network based prediction system changes its input neurons according to a previously detected and recorded match found in a database of anomalous days, in order to accommodate the anomalous day prediction. Results are based on measured power demands recorded in Ireland from domestic smart-meters between 2009-2011, and focus on small scale residential electrical demands of up to 350 kWh. During anomalous days our dynamic prediction approach achieves forecasting results within 3.63% of the real load, down from the 7.37% obtained by the initial prediction algorithm and the 5.41% achieved by standalone re-prediction, without pattern matching.