Simultaneous transmission based communication techniques

Abstract

Any interference between two communicating point is regarded as unwanted noise in conventional communication networks, since it distorts the received signal. In the last two decades, allowing simultaneous transmission and intentionally accepting the interference of signals has been taken into consideration as opposed to the conventional perspective. Joint source-channel coding and the computation codes are one of the first paradigms that allow simultaneous transmission at the same time and frequency and benefit from it. Joint source-channel coding later inspired many other fundamental work from network coding to consensus algorithms, from distributed detection applications to the emerging machine learning. In this thesis, the simultaneous transmission techniques are investigated and two security methods are proposed based on the simultaneous transmission. The chapters of this thesis can be placed under two titles. In the first part, the readers will find a thorough map of the wireless communication literature that benefits from the superposition of signals. The studies are grouped together depending on their purpose and application area. Later, they are presented along with the details such as their contributions and performance metrics. These studies show that the simultaneous transmission can bring scalability, security, low-latency, low-complexity and energy efficiency for certain wireless distributed networks. This part of the thesis emphasizes how the physical layer can be beneficial for unconventional network structures. Especially, the internet of things (IoT) requires unconventional designs to support large networks where the nodes are often resource limited. Simultaneous transmission-based techniques show great opportunities in these scenarios. Also, the attention on the simultaneous transmission applications is expected to grow larger since the application areas of the IoT are constantly spreading. In the second part, an authentication method and a key generation method is presented for multi-user networks. These methods are inspired by the analog function computation (AFC) studies that reduce the computational load of the receiver by computing functions over the air. In essence, AFC methods use signal processing at the transmitters and receivers to invert the fading effect such that the channel model matches with summation operation. After this point, the signal processing is extended to compute any other function. However, existing AFC methods always lose the individual channel inputs over the channel and the receiver only obtains a function of these inputs. In the authentication method, a novel signal processing is developed to harness the individual data from the channel and authenticate it against spoofers. For this purpose, Gaussian prime integers are involved in the signal processing. The pre-processing function encodes the messages at the signal amplitudes in the logarithmic forms. Also the messages are encoded as the exponents of the Gaussian prime "identifiers" which is unique to each user. These prime identifiers enable the extraction of individual data from the superimposed signals. Gaussian primes and the fading ensure the detection of spoofing attacks. For a successful detection, the spoofer is assumed to have larger channel estimation error than the legitimate users. The symbol error rate (SER) and the receiver operating characteristics (ROC) are investigated with simulations in order to verify the feasibility of the proposed approach. In the key generation method, time and bandwidth efficiency of multi-user networks are improved and their dependency on a center node (or an external third party) is removed. Similar to the authentication method, channel model is adjusted with pre and post processing functions. However in this model, the channel is matched with a key generating function that outputs a shared key at each node. The key is assumed to be the combination of multiple key components that are taken from each node. Half duplex and full duplex communication scenarios are considered for the key generation purpose. In the half-duplex scenario, key components are pre-processed and simultaneously transmitted. The pre-processing functions make sure that only the targeted receiver obtains the meaningful information after post-processing. Repeating this process for every node completes the key generation process without leaking the key to the eavesdroppers. This scenario especially removes the dependency on a center node which is required in many traditional method to distribute prior information to each node. This scenario is investigated with simulations and the error probability results are presented. Moreover, the full-duplex scenario is considered where each node simultaneously transmits and receives key components. This model manages to provide a secret key to multiple nodes in a single communication.