Distributed multi-task learning (MTL) is a learning paradigm where distributed users simultaneously learn multiple tasks by leveraging the correlations among tasks. However, distributed MTL suffers ...from a more severe communication bottleneck than single-task learning as more than one models need to be transmitted in the communication phase. To address this issue, we investigate the hierarchical MTL system where distributed users wish to jointly learn different learning models orchestrated by a central server with the help of multiple relays. We propose a coded distributed computing scheme for hierarchical MTL systems that jointly exploits the network topology and relays' computing capability to create coded multicast opportunities to improve communication efficiency. We theoretically prove that the proposed scheme can significantly reduce the communication loads both in the uplink and downlink transmissions between relays and the server. To further illustrate the optimality of the proposed scheme, we derive information-theoretic lower bounds on the minimum uplink and downlink communication loads and prove that the gaps between achievable upper bounds and lower bounds are within the minimum number of connected users among all relays. In particular, when the network topology can be delicately designed, the proposed scheme can achieve the information-theoretic optimal communication loads. Experiments on real-world datasets show that our proposed scheme can greatly reduce the overall training time compared to the conventional hierarchical MTL scheme.
Coded caching scheme provides us an effective framework to realize additional coded multicasting gain by exploiting coding into multiple transmitted signals. The goal of coded caching design is to ...jointly optimize the placement and delivery scheme so as to minimize the amount of transmitted coded signals. However, few research efforts consider multiple-round coded caching design problem, in which fixed and mobile users may coexist in one network, namely different number of active users present in successive rounds. Obviously, such dynamic network configurations may lead to undesired frequent placement caching content updating at user sides, if we assume coded caching scheme for all users in each round separately. Thus how to tailor the coded caching design, such that the frequent caching content updating in the placement phase can be avoided, and simultaneously retaining full caching gain in delivery phase in multiple rounds will become highly desirable. In this paper, by carefully examining the bipartite graph representation of the coded caching scheme, a dynamic centralized coded caching design is proposed on the basis of the concatenating-based placement and the saturating matching based delivery scheme. Our analysis unveils that, the proposed dynamic coded caching scheme can realize the flexible coded multicast, and is order optimal.
This letter considers an extension of multi-access coded caching networks referred to as the multi-access multi-input single-output (MAMISO) networks, where a server with multiple transmit antennas ...is connected to a number of single receive antenna users over a broadcast link and each user can access multiple neighboring cache nodes in a cyclic wrap-around fashion. For such networks, we propose a novel coded caching scheme based on an effective transform over any regular placement delivery array (PDA). By a delicate design of the cache placement, the resulting scheme can yield a degrees of freedom (DoF) as large as possible. It is shown that our proposed scheme has advantage either in the DoF or the subpacketization size over the existing schemes under some parameters.
This paper considers a hierarchical caching system where a server connects with multiple mirror sites, each connecting with a distinct set of users, and both mirror sites and users are equipped with ...caching memories. Although there already exist works studying this setup and proposing coded caching schemes to reduce transmission loads, two main problems are remained to address: 1) the optimal communication load R 1 under the uncoded placement for the first layer is still unknown. 2) the previous schemes are based on Maddah-Ali and Niesen's data placement and delivery, which require high subpacketization level. How to achieve a good tradeoff between transmission loads and subpacketization level for the hierarchical caching system is unclear. In this paper, we aim to address these two problems. We first propose a new combination structure named hierarchical placement delivery array (HPDA), which characterizes the data placement and delivery for a hierarchical caching system. Then we construct two classes of HPDAs, where the first class leads to a scheme achieving the optimal R 1 for some cases, and the second class requires a smaller subpacketization level at the cost of slight increase in transmission loads.
Optimal frequency-hopping sequences can be constructed from cyclic (
υ, K
, 1)-PBDs. A (
υ, K
, 1) difference family can be used to construct a (
υ, K
, 1)-PBD. A lot of work has been done on the ...existence of (
υ, k
, 1) difference families when
K
= {
k
} is a singleton. For
K
= {3, 4}, it is proved that for each prime power
υ
≡ 1 (mod 18) and
υ
⩾ 19, there exists a balanced (
υ, K
, 1) difference family. In this paper, it is proved that there exists a balanced (
υ
, {3, 6}, 1) difference family for each prime power
υ
≡ 1 (mod 36) and
υ
⩾ 37. An infinite class of optimal frequency-hopping sequence is then obtained.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
Adaptive network coded cooperation (ANCC) provides a general coded cooperation framework to map the communication connections of a wireless network into the coding constraint. However, random ...selection policy is assumed at each node to select its retrieval subset, which cannot avoid illness in code graph at the sink. In this paper, the concept of indicator matrix is proposed to regulate the retrieval subset selection to derive the enhanced ANCC (EANCC) scheme. It is unveiled that the indicator matrix based retrieval subset selection policy can effectively tailor the girth and degree characteristics of the resultant code graph. Since the most harmful short cycles in the network coded cooperation can be completely eliminated, the proposed EANCC approach promises better error performance than the existing ANCC scheme. Numerical results show that it performs 13 dB better than the ANCC scheme at the bit error rate performance of <inline-formula><tex-math notation="LaTeX">10^{-5}</tex-math></inline-formula> in seven-user network over the Rayleigh block fading channel.
With the development of microprocessor technology, the base stations (BSs) are equipped with more and more powerful storage and computing ability and then can be used as edge servers in the edge ...computing network. As a result, there is an extreme pressure on transmission load in the edge computing network during the peak traffic time. Coded caching is regarded as an efficient technology to reduce the transmission load. In this paper, we focus on designing the coded caching scheme for any fixed subpacketization and dynamic number of users by constructing an appropriate matrix called placement delivery array (PDA) based on Maddah-Ali and Niesen (MN) scheme. In order to accommodate the dynamic number of users, we remove some columns from the right side of the well known conjugate MN PDA to obtain a new flexible PDA with dynamic column number. It is worth noting that when the number of columns deleted is in a certain range, the transmission load of the scheme realized by our obtained PDA is order optimal. In addition, when the number of columns deleted satisfies a certain condition, by means of combinatorial method we can obtain an improved PDA which leads to a smaller transmission load.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Variable-weight optical orthogonal code (OOC) was introduced by G-C Yang for multimedia optical CDMA systems with multiple quality of service (QoS) requirement. In this paper, a construction for ...optimal (v, {3, 4}, 1, {s/(s+1), 1/(s+1)})-OOCs is given. For s = 2, it is proved that for each prime v ≡ 1 (mod 24), there exists a (v, {3, 4}, 1, {2/3, 1/3})-OOC. A recursive construction for cyclic difference family is also presented. By using these constructions, a number of new infinite classes of optimal (v, {3, 4}, 1, Q)-OOCs for Q = {1/2, 1/2} and {2/3, 1/3} are constructed.
Frameproof codes (FPCs) are widely studied as they are classic fingerprinting codes that can protect copyrighted materials. The main interests are construction methods and bounds of the number of ...codewords of FPCs for a fixed length when the alphabet size approaches infinity. In this paper, we focus on the upper bound of the size of FPCs when the fixed length is 4 and the strength is 2. We obtain an upper bound
2
q
2
-
2
q
+
7
on the size of a
q
-ary 2-FPC of length 4 for any positive integer
q
>
48
. The best previously well known bound of this type of FPCs is
2
q
2
-
2
, which is due to Blackburn (SIAM J Discret Math 16:499–510,
2003
). Our new upper bound improves the previous upper bound and it is not very far from the current best lower bound
2
q
2
-
4
q
+
3
obtained from the explicit construction due to Chee and Zhang (IEEE Trans Inf Theory 58:5449–5453,
2012
).
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OBVAL, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
PDF
