Ph.D. Dissertation Defense: Abdulrahman Baknina Thursday, May 10, 2018
2:00 p.m. AVW 2168
For More Information: Maria Hoo 301 405 3681 mch@umd.edu

ANNOUNCEMENT: Ph.D. Dissertation Defense

Name: Abdulrahman Baknina

Committee:

Professor Sennur Ulukus, Chair/Advisor

Professor Richard La

Professor Gang Qu

Professor Nuno Martins

Professor Amr Baz, Dean's Representative

Date/ Time: Thursday, May 10, at 2:00 pm

Place: AVW 2168

Title: Energy Harvesting Communication Networks: Online Policies, Temperature Considerations, and Age of Information

Abstract:

This dissertation focuses on characterizing energy management policies for energy harvesting communication networks in the presence of stochastic energy arrivals and temperature constraints. When the energy arrivals are stochastic and are known only causally at the transmitter, we study two performance metrics: throughput and age of information (AoI). When the energy harvesting system performance is affected by the change of the temperature, we consider the throughput metric.

When the energy arrivals are stochastic, we study the throughput maximization problem for several network settings.

We first consider an energy harvesting broadcast channel where a transmitter serves data to two receivers on the downlink. The battery at the transmitter in which the harvested energy is stored is of finite size. We focus on online transmission schemes where the transmitter knows the energy arrivals only causally as they happen. We consider the case of general independent and identically distributed (i.i.d.) energy arrivals, and propose a near-optimal strategy coined fractional power constant cut-off (FPCC) policy. We show that the FPCC policy is near-optimal in that it yields rates that are within a constant gap from the optimal rate region, for all system parameters.

Next, we study online transmission policies for a two-user multiple access channel where both users harvest energy from nature. The energy harvests are i.i.d. over time, but can be arbitrarily correlated between the two users. The transmitters are equipped with arbitrary but finite-sized batteries.

We propose a distributed fractional power (DFP) policy, which users implement distributedly with no knowledge of the other user's energy arrival or battery state.

We show that the proposed DFP is near-optimal as in the broadcast channel case.

Then, we consider online power scheduling for energy harvesting channels in which the users incur processing cost per unit time that they are on. The presence of processing costs forces the users to operate in a bursty mode.

We consider the single-user and two-way channels. For the single-user case, we consider the case of the general i.i.d.~energy arrivals. We propose a near-optimal online policy for this case. We then extend our analysis to the case of two-way energy harvesting channels with processing costs; in this case, the users incur processing costs for being on for transmitting or receiving data. Our proposed policy is distributed, which users can apply independently with no need for cooperation or coordination between them.

Next, we consider a single-user channel in which the transmitter is equipped with finite-sized data and energy buffers. The transmitter receives energy and data packets randomly and intermittently over time and stores them in the finite-sized buffers. The arrival amounts are known only causally as they happen. We focus on the special case when the energy and data arrivals are fully-correlated. We propose a structured policy and bound its performance by a multiplicative gap from the optimal. We then show that this policy is optimal when the energy arrivals dominate the data arrivals, and is near-optimal when the data arrivals dominate the energy arrivals.

Then, we consider another performance metric which captures the freshness of data, i.e., AoI. For this metric, we first consider an energy harvesting transmitter sending status updates to a receiver over an erasure channel. The energy arrivals and the channel erasures are i.i.d. and Bernoulli distributed in each slot.

In order to combat the effects of the erasures in the channel and the uncertainty in the energy arrivals, we use channel coding to encode the status update symbols.

We consider two types of channel coding: maximum distance separable (MDS) codes and rateless erasure codes.

For each of these models, we study two achievable schemes: best-effort and save-and-transmit. We analyze the average AoI under each of these policies. We show that rateless coding with save-and-transmit outperforms all other schemes.

Next, we consider a scenario where the transmitter harvests i.i.d. Bernoulli energy arrivals and status updates carry information about an independent message. The transmitter encodes this message into the timings of the status updates. The receiver needs to extract this encoded information, as well as update the status of the observed phenomenon. The timings of the status updates, therefore, determine both the AoI and the message rate (rate). We study the tradeoff between the achievable message rate and the achievable average AoI. We propose several achievable schemes and compare their rate-AoI performances.

Then, with the motivation to understand the effects of temperature sensitivity on wireless data transmission performance for energy harvesting communication networks, we study several temperature models. We assume non-causal knowledge of the energy arrivals. First, we consider throughput maximization in a single-user energy harvesting communication system under continuous time energy and temperature constraints. We model three main temperature related physical defects in wireless sensors mathematically, and investigate their impact on throughput maximization. Specifically, we consider temperature dependent energy leakage, effects of processing circuit power on temperature, and temperature increases due to the energy harvesting process itself. In each case, we determine the optimum power schedule.

Next, different from the previous work, we consider a discrete time system where transmission power is kept constant in each slot. We consider two models that capture different effects of temperature. In the first model, the temperature is constrained to be below a critical temperature at all time instants; we coin this model as explicit temperature constrained model. We investigate throughput optimal power allocation for multiple energy arrivals under general, as well as temperature and energy limited regimes. In the second model, we consider the effect of the temperature on the channel quality via its influence on additive noise power; we coin this model as implicit temperature constrained model. In this model, the change in the variance of the additive noise due to previous transmissions is non-negligible. In particular, transmitted signals contribute as interference for all subsequent slots and thus affect the signal to interference plus noise ratio (SINR). In this case, we investigate throughput optimal power allocation under general, as well as low and high SINR regimes. Finally, we consider the case in which implicit and explicit temperature constraints are simultaneously active.

Finally, we extend the discrete time explicit temperature constraint model to a multi-user setting. We consider a two-user energy harvesting multiple access channel where the temperatures of the nodes are affected by the electromagnetic waves due to data transmission. We study the optimal power allocations when the temperatures of the nodes are subject to peak temperature constraints, where each node has a different peak temperature requirement and the nodes have different temperature parameters. We study the optimal power allocation in this case and derive sufficient conditions under which the rate region collapses to a single pentagon.