Exploiting SIMD parallelism on dynamically partitioned parallel network coding for P2P systems

Kim, Deokho; Park, Karam; Ro, Won Woo

doi:10.1016/j.compeleceng.2012.02.009

Cited by 3 publications

(4 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, in [26], Nage et al studied the TCP-aware network coding with opportunistic scheduling. In [28], network coding in practical P2P systems was studied. In [28], network coding in practical P2P systems was studied.…”

Section: Related Workmentioning

confidence: 99%

“…In [27], Wang et al proposed a scheme that adaptively selects relaying methods among different network coding policies for maximizing the throughput of multi-rate wireless networks. In [28], network coding in practical P2P systems was studied. For more details, please refer to them and papers cited therein.…”

Section: Related Workmentioning

confidence: 99%

“…Note that minimizing the upper bound of the Lyapunov drift (see (28)) is equivalent to maximizing E ∑ n V n ðtÞð∑ b μ nb ðtÞ À ∑ a μ an ðtÞÞjVðtÞ Â Ã , and after the following conversion:…”

Section: Appendix Amentioning

confidence: 99%

See 2 more Smart Citations

An efficient network‐coding based back‐pressure scheduling algorithm for wireless multi‐hop networks

Jiao

Yan

Zhang

et al. 2016

Int J Communication

View full text Add to dashboard Cite

Back-pressure scheduling has been considered as a promising strategy for resource allocation in wireless multi-hop networks. However, there still exist some problems preventing its wide deployment in practice.One of the problems is its poor end-to-end (E2E) delay performance. In this paper, we study how to effectively use inter-flow network coding to improve E2E delay and also throughput performance of back-pressure scheduling. For this purpose, we propose an efficient network coding based back-pressure algorithm (NBP), and accordingly design detailed procedure regarding how to consider coding gain in back-pressure based weight calculation and how to integrate it into next hop decision making in the NBP algorithm. We theoretically prove that NBP can stabilize the networks. Simulation results demonstrate that NBP can not only improve the delay performance of back-pressure algorithm, but also achieve higher network throughput.To ease the understanding of how the back-pressure algorithm in [1] works, here we present a brief introduction. Consider that time is slotted. Each network node maintains a per-flow queue for each flow passing through it. For each time slot, the back-pressure algorithm works to schedule an optimal link set, that is a set of non-interference links whose total link-weights can contribute to a global maximum sum, to transmit data packets. Here, the weight of a link (also called link-weight) is assigned as the largest flow-weight on the link, where the weight of a flow (flow-weight) equals the differential of the flow's queue backlogs between the two endpoints of the link. Obviously, to obtain a global optimal schedule set, global network state knowledge is required. It was shown in [11] that the complexity for computing the optimal schedule is O(|V| 3 ) under 1-hop interference model (where |V| represents the number of nodes in the network) and in general NP-Hard under K-hop interference model (K≥2). Recently, much progress has been made in reducing the computational complexities and also designing distributed implementations of back-pressure algorithms. For more details along this direction, please refer to [12][13][14] and references cited therein.Back-pressure based scheduling usually leads to poor E2E delay performance because of the following two reasons. First, routing decisions made by pure back-pressure algorithm often causes unnecessarily long data paths and thus results in large E2E delay. To address this issue, some remedies have been proposed. For example, the Back-pressure Collection Protocol (BCP) in [7] uses a punish factor to avoid routing loop. In [9], Athanasopoulou et al. proposed a packet-bypacket adaptive back-pressure routing and scheduling algorithm PARN. To encourage packets to go along shorter paths, PARN introduces two strategies: (i) calculation of link weight based on backlog differential of shadow packets, whose total amount is (1 + ε) times that of the native packets and are injected into network from the flow sources; and (ii) M-back-pressure such that a link can o...

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

An efficient network‐coding based back‐pressure scheduling algorithm for wireless multi‐hop networks

Jiao

Yan

Zhang

et al. 2016

Int J Communication

View full text Add to dashboard Cite

show abstract

“…However, their dynamic partitioning method produced additional delay because each row received was partitioned at Stage A and Stage E with different methods. Deokho Kim [20] presented a new data manipulation method to utilize SIMD instruction sets which can be successfully integrated into the dynamic partitioning of thread-level workload distribution. Minwoo Kim [12] revealed the parallelized non-progressive network coding with block-wise Gauss-Jordan elimination and tilling algorithm but the process could not start decoding at the receiver until all n linear independent packets are received.…”

Section: Related Workmentioning

confidence: 99%

Semi-progressive Network Coding Algorithm on Multi-core Processor

Zhu¹,

Xu²,

Zhu³

et al. 2015

JCP

View full text Add to dashboard Cite

Network coding is a popular research topic which can help to improve throughput, reliability and security of communications. However, its decoding process is usually time consuming and the delay is sometime significant. Though the progressive network coding with Gauss-Jordan elimination can reduce the decoding time, the workload cannot be allocated equally among all processor cores. Those problems degrade network coding performance. In this paper, we put forward Semi-Progressive network coding, a new algorithm to narrow the workload gap between any two cores on a multi-core processor. We convert the decoding process into solving the system of linear equations. In addition, we propose a task allocation method. The result of theoretical analysis and calculation show that our algorithm can improve the performance and reduce the delay of decoding process.

show abstract

Design and evaluation of random linear network coding Accelerators on FPGAs

Kim

Jeong

et al. 2013

ACM Trans. Embed. Comput. Syst.

View full text Add to dashboard Cite

Network coding is a well-known technique used to enhance network throughput and reliability by applying special coding to data packets. One critical problem in practice, when using the random linear network coding technique, is the high computational overhead. More specifically, using this technique in embedded systems with low computational power might cause serious delays due to the complex Galois field operations and matrix handling. To this end, this article proposes a high-performance decoding logic for random linear network coding using field-programmable gate-array (FPGA) technology. We expect that the inherent reconfigurability of FPGAs will provide sufficient performance as well as programmability to cope with changes in the specification of the coding. The main design motivation was to improve the decoding delay by dividing and parallelizing the entire decoding process. Fast arithmetic operations are achieved by the proposed parallelized GF ALUs, which allow calculations with all the elements of a single row of a matrix to be performed concurrently. To improve the flexibility in the utilization of the FPGA components, two different decoding methods have been designed and compared. The performance of the proposed idea is evaluated by comparing with the performance of the decoding process executed by general-purpose processors through an equivalent software algorithm. Overall, a maximum throughput of 65.98 Mbps is achieved with the proposed FPGA design on an XC5VLX110T Virtex 5 device. In addition, the proposed design provides speedups of up to 13.84 compared to an aggressively parallelized software decoding algorithm run on a quad-core AMD processor. Moreover, the design affords 12 times higher power efficiency in terms of throughput per watt than an ARM Coretex-A9 processor.

show abstract

Exploiting SIMD parallelism on dynamically partitioned parallel network coding for P2P systems

Cited by 3 publications

References 26 publications

An efficient network‐coding based back‐pressure scheduling algorithm for wireless multi‐hop networks

An efficient network‐coding based back‐pressure scheduling algorithm for wireless multi‐hop networks

Semi-progressive Network Coding Algorithm on Multi-core Processor

Design and evaluation of random linear network coding Accelerators on FPGAs

Contact Info

Product

Resources

About