Elvin Isufi

Autoregressive Moving Average Graph Filtering

One of the cornerstones of the field of signal processing on graphs are graph filters, direct analogs of classical filters, but intended for signals defined on graphs. This paper brings forth new insights on the distributed graph filtering problem. We design a family of autoregressive moving average (ARMA) recursions, which are able to approximate any desired graph frequency response, and give exact solutions for specific graph signal denoising and interpolation problems. The philosophy to design the ARMA coefficients independently from the underlying graph renders the ARMA graph filters suitable in static and, particularly, time-varying settings. The latter occur when the graph signal and/or graph topology are changing over time. We show that in case of a time-varying graph signal, our approach extends naturally to a two-dimensional filter, operating concurrently in the graph and regular time domain. We also derive the graph filter behavior, as well as sufficient conditions for filter stability when the graph and signal are time varying. The analytical and numerical results presented in this paper illustrate that ARMA graph filters are practically appealing for static and time-varying settings, as predicted by theoretical derivations.

Separable autoregressive moving average graph-temporal filters

Despite their widespread use for the analysis of graph data, current graph filters are designed for graph signals that do not change over time, and thus they cannot simultaneously process time and graph frequency content in an adequate manner. This work presents ARMA2D, an autoregressive moving average graph-temporal filter that captures jointly the signal variations over the graph and time. By its unique nature, this filter is able to achieve a separable 2-dimensional frequency response, making it possible to approximate the filtering specifications along both the graph and temporal frequency domains. Numerical results show that the proposed solution outperforms the state of the art graph filters when the graph signal is time-varying.

Stochastic graph filtering on time-varying graphs

We have recently seen a surge of work on distributed graph filters, extending classical results to the graph setting. State of the art filters have however only been examined from a deterministic standpoint, ignoring the impact of stochasticity in the computation (e.g., temporal fluctuation of links) and input (e.g., the value of each node is a random process). Initiating the study of stochastic graph signal processing, this paper shows that a prominent class of graph filters, namely autoregressive moving average (ARMA) filters, are suitable for the stochastic setting. In particular, we prove that an ARMA filter that operates on a stochastic signal over a stochastic graph is equivalent, in the mean, to the same filter operating on the expected signal over the expected graph. We also characterize the variance of the output and we provide an upper bound for its average value among different nodes. Our results are validated by numerical simulations.

Advanced flooding-based routing protocols for underwater sensor networks

Flooding-based protocols are a reliable solution to deliver packets in underwater sensor networks. However, these protocols potentially involve all the nodes in the forwarding process. Thus, the performance and energy efficiency are not optimal. In this work, we propose some advances of a flooding-based protocol with the goal to improve the performance and the energy efficiency. The first idea considers the node position information in order to reduce the number of relays that may apply flooding. Second, a network coding-based protocol is proposed in order to make a better use of the duplicates. With network coding, each node in the network recombines a certain number of packets into one or more output packets. This may give good results in flooding-based protocols considering the high amount of packets that are flooded in the network. Finally, a fusion of both ideas is considered in order to exploit the benefits of both of them.

2-Dimensional finite impulse response graph-temporal filters

Finite impulse response (FIR) graph filters play a crucial role in the field of signal processing on graphs. However, when the graph signal is time-varying, the state of the art FIR graph filters do not capture the time variations of the input signal. In this work, we propose an extension of FIR graph filters to capture also the signal variations over time. By considering also the past values of the graph signal, the proposed FIR graph filter extends naturally to a 2-dimensional filter, capturing jointly the signal variations over the graph and time. As a particular case of interest we focus on 2-dimensional separable graph-temporal filters, which can be implemented in a distributed fashion at the price of higher communication costs. This allows us to give filter specifications and perform the design independently in the graph and temporal domain. The work is concluded by analyzing the proposed approach for stochastic graph signals, where the first and second order moments of the output signal are characterized.

Network Coding for Flooding-Based Routing in Underwater Sensor Networks

In this work, we propose a flooding-based routing protocol using network coding for underwater communications. Due to the high amount of duplicates that flooding-based protocols flood into the network, the sharing of information between the duplicates can improve the packet delivery ratio (PDR). Our simulations show that network coding increases the PDR, but a price is paid in terms of end-to-end delay and number of forwarded duplicates, with respect to other flooding-based protocols. In order to reduce the number of duplicates, while keeping the PDR and the end-to-end delay unchanged, we propose to upgrade the protocol with specific geographical information of the nodes.

Distributed recursive least squares strategies for adaptive reconstruction of graph signals

This work proposes distributed recursive least squares (RLS) strategies for adaptive reconstruction and learning of signals defined over graphs. First, we introduce a centralized RLS estimation strategy with probabilistic sampling, and we propose a sparse sensing method that selects the sampling probability at each node in the graph in order to guarantee adaptive signal reconstruction and a target steady-state performance. Then, a distributed RLS strategy is derived and is shown to be convergent to its centralized counterpart. The performed numerical tests show the performance of the proposed adaptive method for distributed learning of graph signals.

Autoregressive moving average graph filters a stable distributed implementation

We present a novel implementation strategy for distributed autoregressive moving average (ARMA) graph filters. Differently from the state of the art implementation, the proposed approach has the following benefits: (i) the designed filter coefficients come with stability guarantees, (ii) the linear convergence time can now be controlled by the filter coefficients, and (iii) the stable filter coefficients that approximate a desired frequency response are optimal in a least squares sense. Numerical results show that the proposed implementation outperforms the state of the art distributed infinite impulse response (IIR) graph filters. Further, even at fixed distributed costs, compared with the popular finite impulse response (FIR) filters, at high orders our method achieves tighter low-pass responses, suggesting that it should be preferable in accuracy-demanding applications.

Distributed sparsified graph filters for denoising and diffusion tasks

Generally in distributed signal processing, and specifically in distributed graph filters, reducing the communication and computational complexity plays a key role in the network lifetime. In this work we present a novel algorithm to sparsify the graph filtering operation in a random way, where each node decides locally with a certain probability with which of its neighbors to communicate. We show that, if the filter coefficients are changed accordingly, the first and second order moment of the stochastic output are identical to the deterministic filter output and bounded, respectively. We apply our idea on the tasks of signal denoising and diffusion. Numerical results show that the distributed implementation costs of the filter can be reduced up to 95% with a variance of 10−3 from the deterministic output.