We propose two coded schemes for the distributed computing problem of multiplying a matrix by a set of vectors. The first scheme is based on partitioning the matrix into submatrices and applying maximum distance separable (MDS) codes to each submatrix. For this scheme, we prove that up to a given number of partitions the communication load and the computational delay (not including the encoding and decoding delay) are identical to those of the scheme recently proposed by Li et al., based on a single, long MDS code. However, due to the use of shorter MDS codes, our scheme yields a significantly lower overall computational delay when the delay incurred by encoding and decoding is also considered. We further propose a second coded scheme based on Luby Transform (LT) codes under inactivation decoding. Interestingly, LT codes may reduce the delay over the partitioned scheme at the expense of an increased communication load. We also consider distributed computing under a deadline and show numerically that the proposed schemes outperform other schemes in the literature, with the LT code-based scheme yielding the best performance for the scenarios considered.

We present two novel federated learning (FL) schemes that mitigate the effect of straggling devices by introducing redundancy on the devices’ data across the network. Compared to other schemes in the literature, which deal with stragglers or device dropouts by ignoring their contribution, the proposed schemes do not suffer from the client drift problem. The first scheme, CodedPaddedFL, mitigates the effect of stragglers while retaining the privacy level of conventional FL. It combines one-time padding for user data privacy with gradient codes to yield straggler resiliency. The second scheme, CodedSecAgg, provides straggler resiliency and robustness against model inversion attacks and is based on Shamir’s secret sharing. We apply CodedPaddedFL and CodedSecAgg to a classification problem. For a scenario with 120 devices, CodedPaddedFL achieves a speed-up factor of 18 for an accuracy of 95% on the MNIST dataset compared to conventional FL. Furthermore, it yields similar performance in terms of latency compared to a recently proposed scheme by Prakash et al . without the shortcoming of additional leakage of private data. CodedSecAgg outperforms the state-of-the-art secure aggregation scheme LightSecAgg by a speed-up factor of 6.6–18.7 for the MNIST dataset for an accuracy of 95%.

We consider a frame asynchronous coded slotted ALOHA (FA-CSA) system for uncoordinated multiple access, where users join the system on a slot-by-slot basis according to a Poisson random process, and in contrast to standard frame synchronous CSA (FS-CSA), users are not frame-synchronized. We analyze the performance of FA-CSA in terms of packet loss rate and delay. In particular, we derive the (approximate) density evolution that characterizes the asymptotic performance of FA-CSA when the frame length goes to infinity. We show that, if the receiver can monitor the system before anyone starts transmitting, a boundary effect similar to that of spatially coupled codes occurs, which greatly improves the iterative decoding threshold. Furthermore, we derive tight approximations of the error floor (EF) for the finite frame length regime, based on the probability of occurrence of the most frequent stopping sets. We show that, in general, FA-CSA provides better performance in both the EF and waterfall regions as compared to FS-CSA. Moreover, FA-CSA exhibits better delay properties than FS-CSA.

We present a framework for the analysis of the error floor of coded slotted ALOHA (CSA) for finite frame lengths over the packet erasure channel. The error floor is caused by stopping sets in the corresponding bipartite graph, whose enumeration is, in general, not a trivial problem. We therefore identify the most dominant stopping sets for the distributions of practical interest. The derived analytical expressions allow us to accurately predict the error floor at low to moderate channel loads and characterize the unequal error protection inherent in CSA.

We study the impact of phase noise on the downlink performance of a multiuser multiple-input-multiple-output (MIMO) system, where the base station (BS) employs a large number of transmit antennas M. We consider a setup where the BS employs M _osc free-running oscillators, and M/M _osc antennas are connected to each oscillator. For this configuration, we analyze the impact of phase noise on the performance of zero forcing (ZF), regularized ZF, and matched filter precoders when M and the number of users K are asymptotically large, whereas the ratio M/K = β is fixed. We analytically show that the impact of phase noise on the signal-to-interference-plus-noise ratio (SINR) can be quantified as an effective reduction in the quality of the channel state information (CSI) available at the BS when compared with a system without phase noise. As a consequence, we observe that as M _osc increases, the SINR performance of all considered precoders degrades. On the other hand, the variance of the random phase variations caused by the BS oscillators reduces with increasing M _osc . Through Monte Carlo simulations, we verify our analytical results and compare the performance of the precoders for different phase noise and channel noise variances. For all considered precoders, we show that when β is small, the performance of the setup where all BS antennas are connected to a single oscillator is superior to that of the setup where each BS antenna has its own oscillator. However, the opposite is true when β is large and the signal-to-noise ratio (SNR) at the users is low.