Computation Offloading for IoT in C-RAN: Optimization and Deep Learning

Abstract

We consider computation offloading for Internet-of-things (IoT) applications in multiple-inputmultiple-output (MIMO) cloud-radio-access-network (C-RAN). Due to the limited battery life and computational capability in the IoT devices (IoTDs), the computational tasks of the IoTDs are offloaded to a MIMO C-RAN, where a MIMO radio resource head (RRH) is connected to a baseband unit (BBU) through a capacity-limited fronthaul link, facilitated by the spatial filtering and uniform scalar quantization. We formulate a computation offloading optimization problem to minimize the total transmit power of the IoTDs while satisfying the latency requirement of the computational tasks, and find that the problem is non-convex. To obtain a feasible solution, firstly the spatial filtering matrix is locally optimized at the MIMO RRH. Subsequently, we leverage the alternating optimization framework for joint optimization on the residual variables at the BBU, where the baseband combiner is obtained in a closed-form, the resource allocation sub-problem is solved through successive inner convexification, and the number of quantization bits is obtained by a line-search method. As a low-complexity approach, we deploy a supervised deep learning method, which is trained with the solutions to our optimization algorithm. Numerical results validate the effectiveness of the proposed algorithm and the deep learning method.

Federated Learning With a Drone Orchestrator: Path Planning for Minimized Staleness

“…In this setting, the works in [18] and [19] investigated the design of a drone trajectory, along with the problem of communications and computational resource allocation in favor of lowering the energy consumption of an Internet of Things (IoT) network. Combining this with the recent works on efficient offloading of the learning for RL, [20], sparked a new demand for drone-aided intensive edge computations. In a common centralized ML implementation, the drone would act as a sink for all the collected data which is then processed, as in the cloudlet design [18].…”

Joint Optimization of Uplink Power and Computational Resources in Mobile Edge Computing-Enabled Cell-Free Massive MIMO

“…Additionally, [28] addresses the energy minimization problem accounting for imperfect channel state information (CSI) in a single-cell MIMO system. In [29], a successive inner convexification framework to minimize the total transmit power of the devices under latency constraints is proposed; the obtained solution is compared to a supervised deep learning approach, as less computationally demanding and thus more suitable for real-time Internet-of-Things applications.…”

“…• We propose a MEC-enabled CF-mMIMO architecture implementing a user-centric approach both from the radio and the computational resource allocation perspective. Unlike prior studies investigating computation-offloading implementations in distributed networks but operating in a centralized fashion [29], [30], [33], our model considers that users' computational tasks can be executed in a distributed fashion both at the MEC servers of the user-centric cluster and at the cloud CPU.…”

“…Lastly, an additional latency contribution (fronthaul latency) is due to the forwarding of the b k bits from all the APs in the set M k to the cloud CPU, over the fronthaul network, which, assuming synchronous transmission across the APs, amounts to 2b k Mξ/C FH , where ξ denotes the number of bits used to quantize both real and imaginary parts of the uplink data signal y, and C FH is the fronthaul capacity of the link between any AP and the cloud CPU, expressed in bit/s. Hence, the computational offloading must fulfill the following latency constraint [29] b where we assume that L k includes any delay related to the signalling between AP and cloud CPU, and the time needed to send the computational output back to the user. Fig.…”

“…This ensures the same number of available computational resources in the coverage area between the two considered setups. Lastly, the latency requirements for the cellular users is L cell k = 0.7 s. Common to both setups, the number of bits needed to transfer the users' computational tasks, {b k }, are uniformly distributed random integers from the interval [1, 10] Mbits, and the number of CPU cycles needed to run the task itself is conventionally set as a linear function of b k , that is w k = α b k , with α = 50 cycles/bit [29].…”