Makarand. M. Jadhav
Optimization and Machine Learning approach to improve 5G Network performance
Bandwidth is presently a limited and valuable resource. Presently, a fibre optic link is connected to access point in an area. This access point gives us channel for communicating with nearby appliances or devices. Further, channel allocation is done at MAC layer and the same is implemented by physical layer, depending upon which channels are free with the network. Thus allocation/selection of best channel for communication with required QoS as well as QoE using Machine learning for high density network and with less coverage area is biggest challenge to enhance network performance. Further, use of multi-objective optimization techniques will utilize/manage network resources effectively.
Publications
Kantilal. B. Kharat
Antennas using Array Configuration for Multiband Applications
For reducing the size of antenna, fractal geometries have been introduced in the design of antenna. Fractal geometries have two common properties: Self similar property, Space filling property. The self-similarity property of certain fractals results in a multiband behavior. While using space filling properties, a fractal makes reduce antenna size. For improving the inherently narrow bandwidth of a microstrip antenna, it is very effective to use an electrically thick substrate. Several studies for feeding the microstrip antenna with a thick substrate have been reported, such as the L-shaped probe and the capacitive probe-fed structure. It should be noted that coaxial feed or probe feed help to achieve impedance matching. A multiband fractal tree antenna is a candidate for use in applications such as WLAN, ISM, IoT, WSN and Bluetooth.
Publications
Harshad. N. Lokhande
Scene Specific Object Identification for Surveillance Applications
The COVID-19 has entirely changed the world. The world is adjusting with new ways of living to overcome this pandemic situation. The tracing of masked and unmasked persons, can give productive data to identify and track the negligent people to avoid the COVID-19 infection. Nowadays, object tracking an identification using deep learning is an emerging area to achieve better accuracy. For anomaly detection of person in video surveillance, different attributes like location, gender, age, face type, cloth colour and cloth type are important. There is need to design and implement algorithms with masked datasets that can detect variations in masks.
Publications
Sunita. P. Deshmukh
Detection and Diagnosis of Bio-Medical Images with Deep Learning
The time taken to diagnose a disease leads to an important role in the treatment of patients. Liver cancer is one of them. Thus, early detection of cancer or any deformity is proven boon in many cases. Therefore, developing new techniques and applications in the biomedical field on image processing like segmentation, compression, and enhancement are still thrust areas. Further, machine as well as deep learning approaches in training and testing of biomedical CT/MRI/Ultrasound images is need of time.
Publications
Sandip Rikame and P.W. Kulkarni, “Digital electronic Weighing Machine operate on solar energy with emergency light”, in International journal of emerging technology and advanced engineering, 2014 (IJETAE), Page no. 162-168, ISSN/ISBN no. 2250-2459, Vol. 4 Issue 7, July 2014. |