Skip to main content
European Commission logo print header

Distributed storage based on sparse-graph codes

Periodic Reporting for period 1 - DISC (Distributed storage based on sparse-graph codes)

Reporting period: 2015-05-01 to 2017-04-30

The proliferation of mobile devices and the surge of a myriad of multimedia applications has resulted in an exponential growth of the mobile data traffic. It is nowadays accepted that current cellular systems cannot face the unabated growing of today’s digital data volume. Next-generation wireless network desperately need new technologies to improve their robustness to failures, storage efficiency, complexity, and cost efficiency, in order to sustain the information revolution of modern societies.
In this context, wireless caching has emerged as a powerful technique to overcome the backhaul bottleneck, by reducing the backhaul rate and the delay in retrieving content from the network. The key idea is to store popular content closer to the end users. Recently, a novel system architecture named femtocaching was proposed. It consists of deploying a number of small base stations (BSs) with large storage capacity, in which content is stored during periods of offpeak traffic. The mobile users can download content from the small BSs, resulting in a higher throughput per user. It was also proposed to store content directly in the mobile devices. Users can then retrieve content from neighbouring devices using device-to-device (D2D) communication or, alternatively, from the serving BS. The data loss in these systems happens if one or multiple storage nodes fail or leave the network, or if the fading channel characteristics make them unavailable.
In both scenarios, content may be stored using an erasure correcting code, which brings gains with respect to uncoded caching. The use of erasure correcting codes establishes an interesting link between distributed caching for content delivery and distributed storage (DS) for reliable data storage. The key difference is that in the wireless network scenario, data can be downloaded from the storage nodes (the small BSs or the mobile devices) but also from a serving macro BS, which has always the content available. Therefore, the reliability requirements in DS for reliable data storage can be relaxed.
The main objective of this project consists in evaluate the advantages of coded distributed caching with respect to the classical scenario where content is always downloaded from the BS.
The major contribution of this project is the analytical computation of the average download delay in a cellular network when content is cached directly in the mobile devices using erasure correcting codes. The challenge was to obtain closed form expressions for the average file download delay that takes into account several important and practical aspects, such as the fact that the considered nodes are mobile devices that can leave the network in the middle of the download process, and that a device that wants to download a file is not guaranteed to find a fixed number of storage nodes in its vicinity. We consider in fact the case where the mobile devices arrive and depart from the network according to a Poisson random process. Another realistic assumption is that the network information is periodically broadcasted to the devices, and not instantaneously available.
Furthermore, we consider the design of large-scale cellular network using distribute caching. We study a network model that is closer to a large-scale network, where the cell is divided into clusters where D2D links can be activated in order to increase the spatial reuse, and hence to reduce the latency. In this context, we consider more practical aspects, such as the inter and intra-cluster interference, the fact that users may request files, of different popularity, from a library of files, and that the D2D link is not ideal. The main results of the project are collected in [5] and [6].
Most previous works in the literature have focused on the cache hit probability and/or the communication cost and assume that the download of content is instantaneous. In [1], the placement of content encoded using a maximum distance separable (MDS) code to small BSs was investigated and it was shown that a careful placement allows to significantly reduce the backhaul rate. In [2], for the scenario where content is stored directly in the mobile devices, the repairing of the lost data when a device storing data leaves the network was considered. Assuming instantaneous repair, the communication cost of data download and repair was investigated. In [3], [4], a repair scheduling where repair is performed periodically was introduced and analytical expressions for the overall communication cost of content download and data repair as a function of the repair interval were derived. Using these expressions, the overall communication cost entailed by storing content using MDS codes, regenerating codes, and locally repairable codes was evaluated in [4] and it was shown that storing content using erasure correcting code can reduce the overall communication cost with respect to the scenario where content is downloaded solely from the BS. For the scenario where content is stored in small BSs, the expected file download delay has been minimized over the cache content placement, assuming that content is cached using ideal MDS rateless codes. However, while the mobile devices are spread randomly over an area, no mobility of the mobile devices is considered in the analysis. Mobility of the devices, assuming a random walk, is then considered in Monte Carlo simulations.
The novelty of project lies in the fact that for the first time the advantage of distributed caching for wireless content delivery has been analysed in terms of latency, that is a performance measure of very practical interest.
The main achieved scientific objective consists in showing that coded distributed caching can greatly improve the performance in terms of content download delay with respect to the case where content is downloaded from the BS, provided that the BS broadcasts the network information frequently enough. Interestingly, the results of the project also show that the performance improves when the length of the code increases. In particular, simple replication is very inefficient and much better performance are achieved using larger codes (of the same rate).

References:
[1] V. Bioglio, F. Gabry, and I. Land, “Optimizing MDS codes for caching at the edge,” in Proc. IEEE Global Commun. Conf. (GLOBECOM), San Diego, CA, 2015.
[2] J. Pääkkönen, C. Hollanti, and O. Tirkkonen, “Device-to-device data storage for mobile cellular systems,” in Proc. IEEE Global Commun. Conf. (GLOBECOM), 2013.
[3] J. Pedersen, A. Graell i Amat, I. Andriyanova, and F. Brännström, “Repair scheduling in wireless distributed storage with D2D communication,” in Proc. IEEE Inf. Theory Work. (ITW), Jeju Island, Korea, 2015.
[4] J. Pedersen, A. Graell i Amat, I. Andriyanova, and F. Brännström, “Distributed storage in mobile wireless networks with device-to-device communication,” IEEE Trans. Commun., vol. 64, no. 11, pp. 4862–4878, Nov. 2016.
[5] A. Piemontese and A. Graell i Amat, “MDS-coded distributed storage for low delay wireless content delivery,” in Proc. Int. Symp. Turbo Codes & Iterative Inform. Proc. (ISTC), Brest, France, Sep. 2016.
[6] A. Piemontese and A. Graell i Amat, “MDS-coded distributed caching for low delay wireless content delivery,” submitted to IEEE Trans. Commun.. Available on arXiv.
Part of a cell with hexagonal clusters
An example of cell where some nodes can store popular content