Data center operators are increasingly focusing on disaggregated computing to enhance resource utilization efficiency by pooling and logically connecting resources such as CPUs, memories, and accelerators. In disaggregated computing, data transfer between resources is fundamental for service execution. Conventional CPU-mediated data transfer often leads to communication bottlenecks due to data concentration. Therefore, direct data communication between heterogeneous computational resources is critical. To efficiently execute services through direct communication between heterogeneous computational resources, efficient allocation of computational resources and determination of routing paths should be done before service execution. In this paper, we propose a resource allocation method that effectively utilizes heterogeneous computational resources in disaggregated computing by modeling the impact of the characteristics of heterogeneous computational resources and parallel processing. Our method can derive the solution of this model in a practical amount of time. Simulation results show that the proposed method could decrease the number of resources by an average of 28% to 51% compared to the conventional method in a heterogeneous disaggregated computing system while meeting the service performance requirements.

Network microbursts, which are sub-millisecond order bursts of traffic, have gathered attention due to causing network delay and packet loss. However, there are two problems in analyzing the causes of microbursts: how to capture the packet included in the microburst and how to specify the flows causing microbursts. To resolve these problems, we propose a field-programmable gate array (FPGA)-based microburst analysis system. This system detects microbursts with dedicated hardware in sub-millisecond time resolution. It can capture only packets before and after microburst detection triggered by detection with a static threshold for whole traffic. In addition, it can specify the flows causing microbursts by detection with a dynamic threshold for each flow. The experimental results show that the proposed system can capture only packets before and after microburst detection and can correctly specify the flow causing microbursts even in a network with fluctuating bandwidth usage in practical traffic conditions on the basis of network trace data in a datacenter. The proposed system is implemented with Intel®PAC with Arria®10 GX FPGA and consumes relatively small amounts of hardware resources: 51 % ALMs, 16 % registers, and 57 % block memories.

This paper describes a novel wavelength aggregation scheme that we propose. The scheme aims to incrementally upgrade time and wavelength division multiplexing passive optical networks (TWDM-PONs), with the aim of enabling future networks to function effectively. Mobile base stations are split into central units/distributed units (CU/DUs) and remote units (RUs). The optical link between the CU/DUs and the RUs is known as mobile fronthaul (MFH). Future networks that rise above fifth generation networks will need to deploy a huge number of RUs. Thus, quite a few optical fibers will be needed. Accordingly, TWDM-PONs enable MFH to be constructed cost effectively. However, mobile systems tend to upgrade their data rate incrementally up to the standardized maximum data rate in response to user traffic demand. Thus, both upgraded and non-upgraded CU/DUs co-exist in a mobile system at the same time. Therefore, TWDM-PONs have to upgrade their maximum data rates to enable the CU/DUs to be upgraded without stopping any transmission from other CU/DUs that were not upgraded in the PON branch. For this purpose, we propose a wavelength-aggregation scheme with a delay-based frame distribution technique that satisfies the MFH latency requirement. Unlike other schemes, this scheme can upgrade TWDM-PON data rates without stopping any other traffic. Numerical simulations demonstrated that the scheme can provide seamless upgrading that satisfies the MFH latency requirement with a minimized number of wavelengths, one that is less than the upper limit defined in the TWDM-PON specification.

In future radio access systems, since the number of mobile base stations will increase to cope with the increasing mobile traffic, the number of mobile fronthaul (MFH) links will also have to increase. To reduce the MFH link cost, MFH networking has been attracting attention. In particular, the use of a time-division multiplexed passive optical network (TDM-PON) makes the MFH link cost effective. On the other hand, a TDM-PON has a huge latency when forwarding uplink traffic. In a typical dynamic bandwidth allocation (DBA) scheme, an optical network unit (ONU) has a very long transmission waiting time, e.g., several milliseconds. This transmission waiting time in the ONU is a critical problem since the latency requirement for the MFH link is very strict, e.g., less than 250 μs defined by the Third Generation Partnership Project. In this paper, we propose a novel statistical DBA scheme taking high-speed traffic fluctuation to reduce the transmission waiting time. Our proposed scheme allocates bandwidth based on the only ratio of the data amount of each distributed unit and detects the pattern of the transmission interval of the burst signal. We show the feasibility with a numerical simulation and experiments.

Time- and wavelength- division multiplexed passive optical network (TWDM-PON) have attracted attention as the next step in relation to optical access systems. A major application of TWDM-PON is expected to be fronthaul streams in the centralized radio access network (C-RAN) architecture of mobile networks. In the 5G era, the upstream traffic that OLT receives becomes highly bursty, because of the variable data rate generated by the functional split between baseband unit and remote radio heads, and the global synchronization of data transmission with time-division duplex (TDD). To flexibly allocate upstream bandwidth via the dynamic wavelength and bandwidth allocation (DWBA) scheme has been a hot research topic for TWDM-PON. However, there is no existing DWBA scheme for TWDM-PON that is designed to satisfy the strict delay requirement for fronthaul with minimum number of active wavelength channels. Therefore, in this paper we propose a novel efficient DWBA algorithm that minimizes active wavelength channels considering the high burstiness of fronthaul data transmission. With the proposed algorithm, the allowable delay and constraint for allocation timing for each ONU is formulated based on the different propagation delay between the OLT and ONUs. Then, the wavelength channel and bandwidth is allocated with the simple allocation algorithm based on the descending sort by propagation delay. It was confirmed through computer simulations that the proposed algorithm can reduce the number of active wavelength by 50% by considering propagation delay.

We propose a scheme for quickly completing a sorted-table search, which is one of most common searches employed in network processing. In the scheme, the range to be searched is narrowed using only a simple calculation. This narrowing is incorporated into conventional search procedures such as binary search. Simulation results show that our proposed approach reduces the number of times element values are read from the table, which results in faster search.

We propose a novel energy-efficient power saving technique coordinating CPEs via ONU. In addition, we first evaluate the proposal in a 10G-EPON system whose ONU sleep mechanism complies with IEEE Std. 1904.1 SIEPON and IEEE 802.3az EEE. It is revealed that the proposal yields ONU power savings of the order of 42 %. Moreover, it reduces the excessive latency by 40 % compared to unsynchronized power saving control.

Verification of the hardware/software functions of a 10-Gbit/s-class large-scale network systems-on-a-chip (NW SoC) requires the use of multiple field programmable gate array (FPGA) devices. We propose two schemes for the efficient mapping of the design data of the NW SoC into FPGA devices. We implemented practical NW SoC design data for FPGA devices, and evaluated the effectiveness of the schemes. The results show that the proposed schemes can reduce the number of wires by 13% and improve the register access cycle time by 28%.

To reduce the power consumption of optical network units (ONUs), we propose a dual buffer architecture and control schemes for short-time shaping and frame loss prevention. The dual-buffer architecture uses a small-capacity buffer to shut down a large capacity buffer, which is always active in a conventional architecture. The short-time shaping scheme minimizes the architecture overhead by cooperative operation with the optical line terminal (OLT). The frame-loss-prevention scheme is a fail-safe scheme for the cooperative operation. Simulations and experiments show that our proposed approach reduces ONU power consumption without frame loss even while the ONU continuously receives traffic such as video.