Latency in the live audio processing chain has become a concern for audio engineers and system designers because significant delays can be perceived and may affect synchronisation of signals, limit interactivity, degrade sound quality and cause acoustic feedback. In recent years, latency problems have become more severe since audio processing has become digitised, high-resolution ADCs and DACs are used, complex processing is performed, and data communication networks are used for audio signal transmission in conjunction with other traffic types. In many live audio applications, latency thresholds are bounded by human perceptions. The applications such as music ensembles and live monitoring require low delay and predictable latency. Current digital audio systems either have difficulties to achieve or have to trade-off latency with other important audio processing functionalities. This thesis investigated the fundamental causes of the latency in a modern digital audio processing system: group delay, buffering delay, and physical propagation delay and their associated system components. By studying the time-critical path of a general audio system, we focus on three main functional blocks that have the significant impact on overall latency; the high-resolution digital filters in sigma-delta based ADC/DAC, the operating system to process low latency audio streams, and the audio networking to transmit audio with flexibility and convergence. In this work, we formed new theory and methods to reduce latency and accurately predict latency for group delay. We proposed new scheduling algorithms for the operating system that is suitable for low latency audio processing. We designed a new system architecture and new protocols to produce deterministic networking components that can contribute the overall timing assurance and predictability of live audio processing. The results are validated by simulations and experimental tests. Also, this bottom-up approach is aligned with the methodology that could solve the timing problem of general cyber-physical systems that require the integration of communication, software and human interactions.

共著者あり。共著者名:Yoshida Toshikazu, Fukushima Kikuro. / Hokkaido University (北海道大学) / 博士 / 医学

The performance of low latency Network on Chip (NoC) architectures, which incorporate fast bypass paths to reduce communication latency, is limited by crosstalk induced skewing of signal transitions on link wires. As a result of crosstalk interactions between wires, signal transitions belonging to the same flit or bit vector arrive at the destination at different times and are likely to violate setup and hold time constraints for the design. This thesis proposes a two-step technique: TransSync- RecSync, to dynamically eliminate packet errors resulting from inter-bit-line transition skew. The proposed approach adds minimally to router complexity and involves no wire overhead. The actual throughput of NoC designs with asynchronous bypass designs is evaluated and the benefits of augmenting such schemes with the proposed design are studied. The TransSync, TransSync-2-lines and RecSync schemes described here are found to improve the average communication latency by 26%, 20% and 38% respectively in a 7X7 mesh NoC with asynchronous bypass channel. This work also evaluates the bit-error ratio (BER) performance of several existing crosstalk avoidance and error correcting schemes and compares them to that of the proposed schemes. Both TransSync and RecSync scheme are dynamic in nature and can be switched on and off on-the-fly. The proposed schemes can therefore be employed to impart unequal error protection (UEP) against intra-flit skewing on NoC links. In the UEP, a larger fraction of the energy budget is spent in providing protection to those parts of the data being transmitted on the link which have a higher priority, while expending smaller effort in protecting relatively less important parts of the data. This allows us to achieve the prescribed level of performance with lower levels of power. The benefits of the presented technique are illustrated using an H.264 video decoder system-on-chip (SoC) employing NoC architecture. We show that for Akyio test streams transmitted over 3mm long link wires, the power consumption can be reduced by as much as 20% at the cost of an acceptable degradation in average peak signal to noise ratio (PSNR) with UEP.

Similar to cancer cells, HIV-infected cells differ from HIV-uninfected cells in that they have altered interferon signaling pathways, the apparent reason for the selectivity of certain oncolytic viruses (OVs). Therefore, it was hypothesized that use of an OV, such as recombinant Maraba virus (MG1), may be a potential approach to eliminate latently-infected cells constituting the HIV reservoir while sparing HIV-uninfected cells. This was studied in U1, ACH-2, OM-10 and J1.1 cells and their respective HIV-uninfected parent cell lines in addition to CD4+CD25-HLADR- cells from HIV-infected individuals on effective antiretroviral therapy. Although MG1 infected and killed latently HIV-infected U1 cells to a greater degree than the HIV-uninfected parent U937 cells, this was not observed in the other HIV-infected cell lines and their respective parent cell lines. Furthermore, results from primary cells suggest that MG1 alone does not appear to eliminate cells which comprise the major HIV reservoir. Challenges of studying the HIV reservoir and priorities for future studies examining the use of OVs as a potential strategy to eliminate the HIV reservoir are discussed.

Mixed Reality (MR) is a technique to blend together the real life with virtual reality. Using this technique it is, for instance, possible for experts to assist persons several miles away to perform tasks by talking and  visually aid them. In this thesis the main issue is to see how the delay in such a system for remote assistance eects the users. A controlled test was carried out with 20 test persons of dierent backgrounds. The study shows that it is very likely to be able to use an MR system for remote assistance even if there is a delay between the user and the expert. As long as they both are aware of the problem and are able to take it easy and do not have to move around too much it is still possible to work with delays to up to 4000 ms. Furthermore, the average time of completion for a task did not increase with the added delay. It was linear, i.e. the task is not more difficult toperform when the instructions are delayed.

In pigs and other mammals, embryonic losses often occur during implantation when the conceptus (embryo plus its extra-embryonic membranes) attaches to the maternal uterine epithelium. Mechanisms controlling this process are not completely understood. Integrins and growth factors are among many molecules likely involved in controlling implantation. Numerous integrins (ITG), including subunits ITGAV (alpha v), ITGB1 (beta 1), ITGB3 (beta 3), and ITGB5 (beta 5), and transforming growth factor betas (TGFBs), in both latent and active forms, are present at the porcine conceptus-maternal interface. TGFBs are released as latent precursors which cannot interact with TGFBRs prior to their activation. Latency associated peptide (LAP), part of the TGFB latent complex, contains an amino acid sequence Arg-Gly-Asp (RGD) that is found in other extracellular matrix molecules and may interact with and signal through integrins. We hypothesize that LAP will bind to and activate ITGAV-containing heterodimers at the conceptus-maternal interface and that these interactions are a functional component of implantation. We also hypothesize that TGFB acting via TGFBRs has critical roles during peri-implantation, and such roles may include promoting conceptus development, survival, and adhesion. Immunofluorescence was used to colocalize TGFB, LAP, and integrins in porcine peri-implantation uterus and conceptus; immunohistochemistry of phosphorylated SMAD2/3 provided evidence of TGFB activity. Affinity chromatography identified cell surface integrins on porcine trophectoderm that are capable of binding LAP. In vivo, intrauterine infusions of LAP with its native RGD site (LAP-RGD) resulted in inhibition of conceptus elongation; LAP-RGE infusions yielded normal-appearing filamentous conceptuses at d13 of pregnancy. At d24, allantois length and fetal weights were greater in gilts which received LAP-RGE infusions compared to controls which received vehicle only. Results provide evidence for 1) active and latent TGFB in porcine conceptus and uterus; 2) receptor-ligand interactions of integrins and LAP; 3) integrin aggregation and potential focal adhesion formation at the conceptus-maternal interface; and 4) TGFB- and/or integrin-associated mechanisms which regulate conceptus elongation and placental and fetal size. Collectively, results suggest that TGFB and integrins are extensively involved in communication at the porcine conceptus-maternal interface, particularly regulating conceptus development, adhesion, and placental and fetal development.

Mobile IPv6 supports host mobility by dynamically changing IP addresses while mobile nodes roaming in the Internet. However, there still exist performance problems during handoffs, such as handoff latency, packet loss. When a mobile node increases its mobility, performance degradation induced by frequent handoffs grows drastically. In this thesis, we propose a dual-tunnel with buffering (DTWB) mechanism to reduce packet loss ratio during multiple handoffs. Packet buffering at access routers is initiated by mobile nodes when the received signal strength goes below a predefined threshold. The buffered packets are forwarded through dual tunnels, of which the first tunnel is established between the old access router and the new access router, and the second tunnel is established between the new access router and the mobile nodes. For the purpose of evaluation, we perform experiments on NS-2 simulation. The simulation results demonstrate that our proposed mechanism can minimize the packet loss ratio and increase the throughput during multiple handoffs.

Floating-point multipliers are typically power hungry which is undesirable in many embedded applications. This paper proposes a variable-latency floating-point multiplier architecture, which is suitable for low-power, high-performance, and high-accuracy applications. The architecture splits the significand multiplier into upper and lower parts, and predicts the required significand product and sticky bit from upper part. In the case of correct prediction, the computation of lower part is disabled and the rounding operation is significantly simplified so that floating-point multiplication can be completed early. Finally, detailed design and simulation of the floating-point multiplier is presented, together with its evaluation by comparing power consumption with the fast and conventional floating-point multipliers. Experimental results demonstrate that the proposed double-precision multiplier consumes up to 26.41% and 24.97% less power and energy than the fast floating-point multiplier respectively at the expense of only small area and delay overhead. In addition, the results also show that the performance of proposed floating-point multiplier is very approximate to that of fast floating-point multipliers.

Human cytomegalovirus (HCMV), a beta herpesvirus, persists indefinitely in the human host through a life-long, latent infection. HCMV is associated with life threatening pathologies in the immune naïve or compromised and, therefore, understanding of the mechanisms of viral persistence is imperative to human health. The ULb' region of the HCMV genome is selectively lost in high-passage strains of the virus, yet retained in low-passage strains. As such, the ULb' is hypothesized to play a role in immune evasion, pathogenesis, latency, and dissemination. ULb' encoded viral products are poorly characterized, hindering a mechanistic understanding of HCMV persistence. We previously defined a 3.6-kb locus spanning UL133-UL138 within the ULb' region important to viral latency. UL136 is expressed as five protein isoforms ranging from 33-kDa to 19-kDa, arising from alternative transcription and translation mechanisms. We mapped the origins of each isoform through advanced bacterial artificial chromosome recombineering, where each ATG was disrupted and the resulting UL136 recombinant virus was screened for altered expression of the pUL136 isoforms. Remarkably, 8 of the 11 potential translation initiation sites encoded within the ORF are utilized to create the pUL136 isoforms. The pUL136 isoforms have distinct localization and trafficking patterns within the cell, including varying degrees of Golgi association, suggesting each isoform may interface with different cellular components and pathways. Further characterization of UL136 recombinant viruses revealed a complex, antagonistic relationship between the pUL136 isoforms. In endothelial cells, which are important to viral persistence and dissemination due to their ability to maintain a slow, "smoldering" infection, the 33- and 26-kDa isoforms promote replication, while the 25-kDa isoform enhances their combined activity, and the 23-/19-kDa isoforms repress the activity of the 25-kDa isoform. The pUL136 isoforms are also required for virus maturation in endothelial cells, where the 33-kDa is required both for virion envelopment and efficient formation of the perinuclear viral assembly compartment. In both an in vitro CD34⁺ cell culture model of latency and an in vivo NOD-scid IL2Rɣc^(null) humanized mouse model, a virus lacking the 23-/19-kDa isoforms fails to establish latency, instead replicating and disseminating with increased efficiency while viruses lacking the 33- and 26-kDa isoforms fail to efficiently reactivate or disseminate. Our data suggest that the interplay between the pUL136 isoforms maintains an intricate balance of infection that governs replication, latency, and virus dissemination, which ultimately contributes to the role of the UL133/8 locus in mediating outcomes of HCMV infection.

No description available.