INTELLIGENT ALGORITHMS WITH SELECTION OF HYPERPARAMETERS FOR E-HEALTH APPLICATIONS POWERED BY 5G WIRELESS NETWORKS
INTELLIGENT ALGORITHMS WITH SELECTION OF HYPERPARAMETERS FOR E-HEALTH APPLICATIONS POWERED BY 5G WIRELESS NETWORKS

Summary/Abstract: Different from previous generations, the 5G networks has new capabilities due to service-based architecture model and virtualization. The successful broadband networks must be able to handle the growth in the data traffic. The e-Health networks has additional issues as the continuous monitoring of patients suffering from chronic diseases (non-communicable diseases). The wearable devices used for monitoring are supposed to be used for balneo-physio-kinetotherapy (including the body gait index calculation) in the future and this will require and increasing traffic as users and data for 5G networks. Medical data and biomedical data are usually very large (especially for medical images) and the traffic can be critical in some situation when in order to take a decision due to alarms from generated by medical emergency when the data should be provided very fast to the physicians (hospitals, or clinics). An architecture for smart e-Health monitoring including the management of big database open the opportunity to use intelligent algorithm for complex problems, machine learning and artificial intelligence. The possibility to use of three algorithms in simulation and simulators for e-Health 5G wireless network is investigate in this paper. One of the key requirement is low energy consumption due to number of antenna elements at the access points and number of user terminals. The problem optimization address to a mix agglomeration: dense urban area along with a set of dispersed locations in a rural area. The network planning is defined as optimization problem of configuration that depends on BS (Base Station) location and transmission power but as novelty, the constraints due to inclusion of rural area are also included in feasible solution. The constraints refer to two situations: the relief (that can be natural zone) or imposed black zone due external factors. Three algorithms are examined: Realcoded Genetic Algorithm for Variable Population - RCGAV), NSGA - II and Gossip, applied to modelling and optimization of power consumption in wireless access networks. Scenarios with simulation of the traffic between the client and the server are taken in to account using known models of distribution: Poisson, Pareto, and Weibull.

• Details
• Contents