The Seer

Our Work

The Seer is focused on streamlining the process of location estimation to benefit telecommunications, assisted GPS (AGPS), and radio enthusiasts. By implementing a neural network that utilizes the received signal data from an array of antennas, our system can create an accurate and adaptably complex model of the environment it is trained in. Analyzing the amplitude and phase of these received signals provides useful data for the creation of our deep learning model. Use of a neural network allows for a model to be created that matches the complexity of the urban indoor environments we are targeting with our prototype. Other methods of pinpointing the location of an incoming signal use models that assume an isotropic environment, while true indoor urban environments are by no means isotropic. Our system is capable of improving the current methods for determining the direction-of-arrival of low-band 5G signals, bolstering communication between a base station and transmitter, while lowering the economic impact of these large scale 5G systems.

Problem Statement

Methods commonly used to find DOA assume an isotropic environment and do not account for: multipath, constructive/destructive interference, non-proportional relationships between power and distance, changes within the environment, etc. This can mean poor performance in complex environments where there are many obstacles and contributors to the EM field. With a non-isotropic environment, such as a cellular consumer's home, the solution to the inverse problem of finding the DOA using the signal parameters can be extremely difficult, increasing the challenges of modeling the propagation medium.

Proposed Solution

We proposed the implementation of a system that uses deep learning to determine the direction-of-arrival of an incoming low-band 5G signal. The system can work well indoors since the environment need not be isotropic. Our neural network is able to develop a model that accounts for the complexity of the environment through training. Our system focuses on sub-6 5G NR location estimation within the 600 MHz (n5) to 850 MHz (n71) band. This project is a proof of concept prototype focused on the static Tx case, able to solve the aforementioned problem by mapping the EM spectrum within the testing environment and developing a model specific to this environment. Through training, the system can be utilized in any environment where low-band 5G signals are present.

Value Proposition

Our system benefits consumers looking to speed up cellular connectivity within their own home or neighborhood. By mapping the EM environment, our system is able to connect a transmitting cell phone with a local base station quickly and accurately. Our system speeds up the pairing process, provides specific information regarding the location of the consumers cell phone (Tx), and reduces unnecessary power expenditure during the connection process. The benefits of our system include increased efficiency, higher accuracy for applications that rely on precise location services such as assisted GPS (AGPS), and optimized power use within the telecommunication system it is utilized in.

Engineering & Marketing Requirements:

MR1): The system must streamline the process of the Rx determining the direction-of-arrival of the incoming signal (less time and less energy than triangulation).
MR2): The system must be able to determine direction-of-arrival within an acceptable range.
MR3): The system can be modified for other environments through training of the neural network.
MR4): The system must be able to handle noise up to a certain threshold.
MR5): The system must be able to understand and work with low-band 5G signals.
MR6): The project should have a interface where the user can see data clearly.
MR7): The system must be in-expensive enough for mass production.

ER1): The system must be able to guess the direction-of-arrival of the transmitted signal with an accuracy level of 90% or greater.
ER2): The system must work for 5G signals transmitted within a radial distance of at least 6 meters.
ER3): The system must be able to come up with a valid model for any environment it is trained in.
ER4): The system must be accurate in the presence of <= -40 dBW of noise.
ER5): The system must be able to work with frequencies in the 600-850 MHz band.
ER6): There must exist a GUI that displays real and accurate data within 60 seconds.
ER7): Prototype must cost less than $600.

Team Straw Hats

Tate Harsch-Hudspeth

Evan Peelen

Victor Madrid

Our Work

Problem Statement

Proposed Solution

Value Proposition

Engineering & Marketing Requirements:

Project Resources: