Computation of Traffic Time Series for Large Populations of IoT Devices.

The Internet of Things (IoT) contains sets of hundreds of thousands of network-enabled devices communicating with central controlling nodes or information collectors. The correct behaviour of these devices can be monitored by inspecting the traffic that they create. This passive monitoring methodology allows the detection of device failures or security breaches. However, the creation of hundreds of thousands of traffic time series in real time is not achievable without highly optimised algorithms. We herein compare three algorithms for time-series extraction from traffic captured in real time. We demonstrate how a single-core central processing unit (CPU) can extract more than three bidirectional traffic time series for each one of more than 20,000 IoT devices in real time using the algorithm DStries with recursive search. This proposal also enables the fast reconfiguration of the analysis computer when new IoT devices are added to the network.

Bio-inspired computational memory model of the Hippocampus: An approach to a neuromorphic spike-based Content-Addressable Memory.

The brain has computational capabilities that surpass those of modern systems, being able to solve complex problems efficiently in a simple way. Neuromorphic engineering aims to mimic biology in order to develop new systems capable of incorporating such capabilities. Bio-inspired learning systems continue to be a challenge that must be solved, and much work needs to be done in this regard. Among all brain regions, the hippocampus stands out as an autoassociative short-term memory with the capacity to learn and recall memories from any fragment of them. These characteristics make the hippocampus an ideal candidate for developing bio-inspired learning systems that, in addition, resemble content-addressable memories. Therefore, in this work we propose a bio-inspired spiking content-addressable memory model based on the CA3 region of the hippocampus with the ability to learn, forget and recall memories, both orthogonal and non-orthogonal, from any fragment of them. The model was implemented on the SpiNNaker hardware platform using Spiking Neural Networks. A set of experiments based on functional, stress and applicability tests were performed to demonstrate its correct functioning. This work presents the first hardware implementation of a fully-functional bio-inspired spiking hippocampal content-addressable memory model, paving the way for the development of future more complex neuromorphic systems.

Remote access protocols for Desktop-as-a-Service solutions.

The use of remote desktop services on virtualized machines is a general trend to reduce the cost of desktop seats. Instead of assigning a physical machine with its operating system and software to each user, it is considerably easier to manage a light client machine that connects to a server where the instance of the user's desktop machine actually executes. Citrix and VMware have been major suppliers of these systems in private clouds. Desktop-as-a-Service solutions such as Amazon WorkSpaces offer a similar functionality, yet in a public cloud environment. In this paper, we review the main offerings of remote desktop protocols for a cloud deployment. We evaluate the necessary network resources using a traffic model based on self-similar processes. We also evaluate the quality of experience perceived by the user, in terms of image quality and interactivity, providing values of Mean Opinion Score (MOS). The results confirm that the type of application running on the remote servers and the mix of users must be considered to determine the bandwidth requirements. Applications such as web browsing result in unexpectedly high traffic rates and long bursts, more than the case of desktop video playing, because the on-page animations are rendered on the server.