Flight delays impact airlines, airports and passengers. Delay prediction is crucial during the decision-making process for all players in commercial aviation, and in particular for airlines to meet their on-time performance objectives. Although many machine learning approaches have been experimented with, they fail in (i) predicting delays in minutes with low errors (less than 15 minutes), (ii) being applied to small carriers i.e., low cost companies characterized by a small amount of data. This work presents a Long Short-Term Memory (LSTM) approach to predicting flight delay, modeled as a sequence of flights across multiple airports for a particular aircraft throughout the day. We then suggest a transfer learning approach between heterogeneous feature spaces to train a prediction model for a given smaller airline using the data from another larger airline. Our approach is demonstrated to be robust and accurate for low cost airlines in Europe.

In this paper, we consider the design of forward error correction tailored specifically for the low end-to-end latency requirements in 5G networks. We present experimental results that highlight a number of issues with conventional approaches and then introduce a new low delay code construction that achieves a superior throughput-delay tradeoff. We analyze its performance both mathematically and experimentally. The mathematical analysis of throughput and delay requires the development of a number of novel analytic tools based on both queuing and coding theories. We implement the low delay code and evaluate its performance in an experimental test bed.

High-performance real-time applications, expected to be of importance in the upcoming 5G era, such as virtual and augmented reality or tele-presence, have stringent requirements on throughput and per-packet in-order delivery delay. Use of multipath transport is gaining momentum for supporting these applications. However, building an efficient, low latency multipath transfer mechanism remains highly challenging. The primary reason for this is that the delivery delay along each path is typically uncertain and time-varying. When the transmitter ignores the stochastic nature of the path delays, then packets sent along different paths frequently arrive out of order and need to be buffered at the receiver to allow in-order delivery to the application. In this paper, we propose Stochastic Earliest Delivery Path First (S-EDPF), a generalization of EDPF which takes into account uncertainty and time-variation in path delays yet has low-complexity suited to practical implementation. Moreover, we integrate a novel low-delay Forward Error Correction (FEC) scheme into S-EDPF in a principled manner by deriving the optimal schedule for coded packets across multiple paths. Finally, we demonstrate, both analytically and empirically, that S-EDPF is effective at mitigating the delay impact of reordering and loss in multipath transport protocols, offering substantial performance gains over the state of the art.

We introduce an optimisation problem to maximise the throughput of class of rateless codes subject to a delay constraint for a point to point erasure packet channel with delayed feedback. We show that this problem is convex. Using the formulation for point to point communication, we extend our approach to the communication of multiple unicast flows in a network.

In this paper, we consider fair privacy in a shared network subject to traffic analysis attacks by an eavesdropper. We initiate the study of the joint trade-off between privacy, throughput and delay in such a shared network as a utility fairness problem and derive the proportional fair rate allocation for networks of flows subject to privacy constraints and delay deadlines.

Polar coding is a recent channel coding technique invented by Arikan to achieve the 'symmetric capacity' of binary-input memoryless channels. Subsequently it was observed by Korada and Urbanke that such codes are also good for lossy channel coding, achieving the `symmetric rate distortion' bound, when the representation alphabet is binary. In this note we extend this result to the case when the representation alphabet is q-ary, for q a prime number.