Search

We introduce a novel category of set functions called Extended Deep Submodular functions (EDSFs), which are neural network-representable. EDSFs serve as an extension of Deep Submodular Functions (DSFs), inheriting crucial properties from DSFs while addressing innate limitations. It is known that DSFs can represent a limiting subset of submodular functions. In contrast, through an analysis of polymatroid properties, we establish that EDSFs possess the capability to represent all monotone submodular functions, a notable enhancement compared to DSFs. Furthermore, our findings demonstrate that EDSFs can represent any monotone set function, indicating the family of EDSFs is equivalent to the family of all monotone set functions. Additionally, we prove that EDSFs maintain the concavity inherent in DSFs when the components of the input vector are non-negative real numbers-an essential feature in certain combinatorial optimization problems. Through extensive experiments, we illustrate that EDSFs exhibit significantly lower empirical generalization error than DSFs in the learning of coverage functions. This suggests that EDSFs present a promising advancement in the representation and learning of set functions with improved generalization capabilities.

We study joint optimization of service placement, request routing, and CPU sizing in a cooperative MEC system. The problem is considered from the perspective of the service provider (SP), which delivers heterogeneous MEC-enabled delay-sensitive services, and needs to pay for the used resources to the mobile network operators and the cloud provider, while earning revenue from the served requests. We formulate the problem of maximizing the SP's total profit subject to the computation, storage, and communication constraints of each edge node and end-to-end delay requirements of the services as a mixed-integer non-convex optimization problem, and prove it to be NP-hard. To tackle the challenges in solving the problem, we first introduce a design trade-off parameter for different delay requirements of each service, which maintains flexibility in prioritizing them, and transform the original optimization problem by the new delay constraints. Then, by exploiting a hidden convexity, we reformulate the delay constraints into an equivalent form. Next, to handle the challenge of the complicating (integer) variables, using primal decomposition, we decompose the problem into an equivalent form of master and inner sub-problems over the mixed and real variables, respectively. We then employ a cutting-plane approach for building up adequate representations of the extremal value of the inner problem as a function of the complicating variables and the set of values of the complicating variables for which the inner problem is feasible. Finally, we propose a solution strategy based on generalized Benders decomposition and prove its convergence to the optimal solution within a limited number of iterations. Extensive simulation results demonstrate that the proposed scheme significantly outperforms the existing mechanisms in terms of the SP's profit, cache hit ratio, running time, and end-to-end delay.

Compute-and-forward is a promising strategy to tackle interference and obtain high rates between the transmitting users in a wireless network. However, the quality of the wireless channels between the users substantially limits the achievable computation rate in such systems. In this paper, we introduce the idea of using intelligent reflecting surfaces (IRSs) to enhance the computing capability of the compute-and-forward systems. For this purpose, we consider a multiple access channel(MAC) where a number of users aim to send data to a base station (BS) in a wireless network, where the BS is interested in decoding a linear combination of the data from different users in the corresponding finite field. Considering the compute-and-forward framework, we show that through carefully designing the IRS parameters, such a scenario's computation rate can be significantly improved. More specifically, we formulate an optimization problem which aims to maximize the computation rate of the system through optimizing the IRS phase shift parameters. We then propose an alternating optimization (AO) approach to solve the formulated problem with low complexity. Finally, via various numerical results, we demonstrate the effectiveness of the IRS technology for enhancing the performance of the compute-and-forward systems, which indicates its great potential for future wireless networks with massive computation requirements, such as 6G.

In this paper, we propose a method of distributed stochastic gradient descent (SGD), with low communication load and computational complexity, and still fast convergence. To reduce the communication load, at each iteration of the algorithm, the worker nodes calculate and communicate some scalers, that are the directional derivatives of the sample functions in some \emph{pre-shared directions}. However, to maintain accuracy, after every specific number of iterations, they communicate the vectors of stochastic gradients. To reduce the computational complexity in each iteration, the worker nodes approximate the directional derivatives with zeroth-order stochastic gradient estimation, by performing just two function evaluations rather than computing a first-order gradient vector. The proposed method highly improves the convergence rate of the zeroth-order methods, guaranteeing order-wise faster convergence. Moreover, compared to the famous communication-efficient methods of model averaging (that perform local model updates and periodic communication of the gradients to synchronize the local models), we prove that for the general class of non-convex stochastic problems and with reasonable choice of parameters, the proposed method guarantees the same orders of communication load and convergence rate, while having order-wise less computational complexity. Experimental results on various learning problems in neural networks applications demonstrate the effectiveness of the proposed approach compared to various state-of-the-art distributed SGD methods.

In this paper, we consider the problem of stochastic optimization, where the objective function is in terms of the expectation of a (possibly non-convex) cost function that is parametrized by a random variable. While the convergence speed is critical for many emerging applications, most existing stochastic optimization methods suffer from slow convergence. Furthermore, the emerging technology of parallel computing has motivated an increasing demand for designing new stochastic optimization schemes that can handle parallel optimization for implementation in distributed systems. We propose a fast parallel stochastic optimization framework that can solve a large class of possibly non-convex stochastic optimization problems that may arise in applications with multi-agent systems. In the proposed method, each agent updates its control variable in parallel, by solving a convex quadratic subproblem independently. The convergence of the proposed method to the optimal solution for convex problems and to a stationary point for general non-convex problems is established. The proposed algorithm can be applied to solve a large class of optimization problems arising in important applications from various fields, such as machine learning and wireless networks. As a representative application of our proposed stochastic optimization framework, we focus on large-scale support vector machines and demonstrate how our algorithm can efficiently solve this problem, especially in modern applications with huge datasets. Using popular real-world datasets, we present experimental results to demonstrate the merits of our proposed framework by comparing its performance to the state-of-the-art in the literature. Numerical results show that the proposed method can significantly outperform the state-of-the-art methods in terms of the convergence speed while having the same or lower complexity and storage requirement.

Wireless power transfer (WPT) is a viable source of energy for wirelessly powered communication networks (WPCNs). In this paper, we first consider WPT from an energy access point (E-AP) to multiple energy receivers (E-Rs) to obtain the optimal policy that maximizes the WPT efficiency. For this purpose, we formulate the problem of maximizing the total average received power of the E-Rs subject to the average and peak power level constraints of the E-AP. The formulated problem is a non-convex stochastic optimization problem. Using some stochastic optimization techniques, we tackle the challenges of this problem and derive a closed-form expression for the optimal solution, which requires the explicit knowledge of the distribution of channel state information (CSI) in the network. We then propose a near-optimal algorithm that does not require any explicit knowledge of the CSI distribution and prove that the proposed algorithm attains a near-optimal solution within a guaranteed gap to the optimal solution. We next consider fairness among the E-Rs and propose a quality of service (QoS) aware fair policy that maximizes a generic network utility function while guaranteeing the required QoS of each E-R. Finally, we study a practical wirelessly powered communication scenario in which the E-Rs utilize their energy harvested through WPT to transmit information to the E-AP. We optimize the received information at the E-AP under its average and peak transmission power constraints and the fairness constraints of the E-Rs. Numerical results show the significant performance of our proposed solutions compared to the state-of-the-art baselines.

In a wirelessly-powered communication network (WPCN), an energy access point (E-AP) supplies the energy needs of the network nodes through radio frequency wave transmission, and the nodes store their received energy in their batteries for possible data transmission. In this paper, we propose an online control policy for energy transfer from the E-AP to the wireless nodes and for data transfer among the nodes. With our proposed control policy, all data queues of the nodes are stable, while the average energy consumption of the network is shown to be within a bounded gap of the minimum energy required for stabilizing the network. Our proposed policy is designed using a quadratic Lyapunov function to capture the limitations on the energy consumption of the nodes imposed by their battery levels. We show that under the proposed control policy, the backlog level in the data queues and the stored energy level in the batteries fluctuate in small intervals around some constant levels. Consequently, by imposing negligible average data drop rate, the data buffer size and the battery capacity of the nodes can be significantly reduced.

In a wireless powered communication network (WPCN), an energy access point supplies the energy needs of the network nodes through radio frequency wave transmission, and the nodes store the received energy in their batteries for their future data transmission. In this paper, we propose an online stochastic policy that jointly controls energy transmission from the EAP to the nodes and data transfer among the nodes. For this purpose, we first introduce a novel perturbed Lyapunov function to address the limitations on the energy consumption of the nodes imposed by their batteries. Then, using Lyapunov optimization method, we propose a policy which is adaptive to any arbitrary channel statistics in the network. Finally, we provide theoretical analysis for the performance of the proposed policy and show that it stabilizes the network, and the average power consumption of the network under this policy is within a bounded gap of the minimum power level required for stabilizing the network.

Radio frequency wireless power transfer (RF-WPT) is an emerging technology that enables transferring energy from an energy access point (E-AP) to multiple energy receivers (E-Rs), in a wireless manner. In practice, there are some restrictions on the power level or the amount of energy that the E-AP can transfer, which need to be considered in order to determine a proper power transfer policy for the E-AP. In this paper, we formulate the problem of finding the optimal policy for two practical scenarios of powerlimited and energy-limited E-APs. The formulated problems are non-convex stochastic optimization problems that are very challenging to solve. We propose optimal and near-optimal policies for the power transfer of the E-AP to the E-Rs, where the optimal solutions require statistical information of the channel states, while the near-optimal solutions do not require such information and perform well in practice. Furthermore, to ensure fairness among E-Rs, we propose two fair policies, namely Max- Min Fair policy and quality-of-service-aware Proportional Fair policy. MMF policy targets maximizing the minimum received power among the E-Rs, and QPF policy maximizes the total received power of the E-Rs, while guaranteeing the required minimum QoS for each E-R. Various numerical results demonstrate the significant performance of the proposed policies.

Providing backhaul connectivity for macro and pico base stations (BSs) constitutes a significant share of infrastructure costs in future heterogeneous networks (HetNets). To address this issue, the emerging idea of flexible backhaul is proposed. Under this architecture, not all the pico BSs are connected to the backhaul, resulting in a significant reduction in the infrastructure costs. In this regard, pico BSs without backhaul connectivity need to communicate with their nearby BSs in order to have indirect accessibility to the backhaul. This makes the radio resource management (RRM) in such networks more complex and challenging. In this paper, we address the problem of cross-layer RRM in HetNets with flexible backhaul. We formulate this problem as a two-timescale non-convex stochastic optimization which jointly optimizes flow control, routing, interference mitigation and link scheduling in order to maximize a generic network utility. By exploiting a hidden convexity of this non-convex problem, we propose an iterative algorithm which converges to the global optimal solution. The proposed algorithm benefits from low complexity and low signalling, which makes it scalable. Moreover, due to the proposed two-timescale design, it is robust to the backhaul signalling latency as well. Simulation results demonstrate the significant performance gain of the proposed solution over various baselines., Comment: We note that the definition of Subproblem 2 (named as Problem P2 in the paper) was missed in the published version of this paper in IEEE Transactions on Wireless Communications. This definition is denoted by equation (8) in this arXiv version