PhD forum: Not so cooperative caching

PURPOSE: It is well-understood that exercise and diet choice can improve several facets of an individual’s cognitive health, but little is known about the use of exercise prescription for long-term memory loss due to chronic consumption of a “western diet.” Thus, this project examined the prescription of interval-running exercise as a way to mitigate diet-induced long-term memory and cognitive decline. METHODS: Male and female C57BL/6 mice were fed either a high-fat diet (HFD; 60% kcal from fat) or a control diet (CD; 14% kcal from fat) from postnatal day (PND) 21. At PND 56, groups from each diet either remained sedentary or underwent eight weeks of an interval-running paradigm with ten cycles of 60 seconds of work and 30 seconds of rest five days per week. At PND 84, long-term memory was tested using several retention tests in a radial arm maze (RAM). An object placement test (OPT) was also used, where subjects were tasked to remember the locations of two objects. After 24 hours of retention, subjects were measured on their abilities to discriminate between novel and original object placements. After OPT was completed, brain samples were collected to test for brain-derived neurotrophic factor (BNDF; Mouse BDNF ELISA Kit). RESULTS: As a combined function of errors and time, there was no performance difference in RAM between exercised CD and HFD subjects (p = .583). Exercised HFD subjects performed significantly better than sedentary HFD subjects (p < .001). In OPT, exercised HFD (p < .001) and CD (p < .001) subjects investigated the novel object placement significantly more than the original placement, while sedentary HFD (p = .071) and CD (p < .236) subjects showed no difference. While BDNF data are not fully collected, behavioral data suggest a difference between exercised and sedentary subjects. CONCLUSIONS: Exercise effectively mitigated long-term memory and cognitive decline. The lack of difference within exercised subjects matters, as it demonstrates the ability of exercise to improve memory performance compared to sedentary subjects. Investigative differences of novel and original object placements in exercised subjects also demonstrates improved memory and cognitive abilities. Results can be used to formulate personalized treatment plans for individuals suffering memory decline due to chronic HFD consumption. Supported by: University of North Georgia Psychological Sciences PARKS Foundation and UNG CURCA Student-Faculty Collaborative Mini-Grant

As a novel Internet architecture, Named Data Networking (NDN) has been proposed. The router used in NDN has a content caching function. In NDN, A distributed caching approach has scalability, and a cooperative caching approach is effective as improving cache hit rate typically. However, it is difficult to improve cache efficiency for a distributed approach, thus a lot of cooperative caching methods have been proposed to enhance the cache efficiency. In this paper, we propose a distributed cache management method, which is Push-based Traffic-Aware distributed Cache management (P-TAC). P-TAC uses the links to improve cache hit rate by using the links having a margin in a transmission band. The routers balance the caching and the pushing for the whole network benefit. Simulation results showed that P-TAC achieves the reduction of the traffic from outer ISP compared with Least Frequently Used (LFU) method with simple forwarding. Compared with other methods, P-TAC showed high performance at the point of cache hit rate and memory efficiency.

Parallelism is a viable solution to constructing high performance object-oriented database systems. In parallel systems based on a shared-nothing architecture, the database is horizontally declustered across multiple processors, enabling the system to employ multiple processors to speedup the execution time of a query. The placement of objects across the processors has a significant impact on the performance of queries that traverse a few objects. The paper describes and evaluates a greedy algorithm for the placement of objects across the processors of a system. Moreover, it describes two alternative availability strategies and quantifies their performance tradeoff using a trace-driven simulation study. >

Named Data Networking (NDN) has a built-in caching capability that is enabled with the help of its Content Store. Caching in NDN has several benefits, such as reducing overhead on the producer side, avoiding a single point of failure, and reducing network load. The primary caching policy of the NDN architecture is to leave copies everywhere. However, this scheme induces significant cache redundancy. Existing advanced cache techniques either periodically share the entire list of cached content at a node or make a caching decision without knowing the cached content at other nodes in the network. We propose an intelligent cache policy that estimates cache distribution and topology through a collaborative network of NDN routers, guided by Interest and Data packets. Additionally, we develop an NDN simulator to facilitate the testing and evaluation of our machine learning caching strategy. This paper details the cache policy, the design and functionality of our NDN simulator, and the experiments comparing our approach to baseline caching strategies.

Nash equilibria in discrete routing games with convex latency functions

As one of the most popular applications currently running on the Internet, the World Wide Web is of an exponential growth in size, which results in the network congestion and server overloading. Web caching as an important technique aims at reducing network traffic, server load, and user-perceived retrieval delays by replicating popular contents on proxy caches. In our paper, we develop an anycast-based cooperative web proxy caching mechanism. It puts response time, hop counts, self-organized load balancing, adaption, and self-configuration as primary concerns. In our scheme, anycast is used to bring a server “nearest” to a client to improve overall performance, all proxies in a local network consists an anycast group. The object placement and replacement problems are formulated as an optimization problem and the solution is obtained by using an anycast-based cooperative proxy caching algorithm. Some simulation experiments have been conduced to evaluate the proposed scheme in terms of a wide range of performance metrics. Analytical and experimental results show that our algorithm outperforms some existing algorithms in response time, hop counts.

We present a novel multiuser virtual reality (VR) system where the physical world is used as a template for the placement of walls, furniture, and objects in the virtual world so as to create a correspondence in scale, spatial layout, and object placement between the two spaces. Through this association between the real and virtual world, users are able to walk freely while wearing a head-mounted device, avoid obstacles like walls and furniture, interact with each other and with objects just like they would in real life, bringing us closer to the realization of realistic collaborative virtual environments. Preliminary deployment during our lab's semi-annual open house shows that the system has potential to offer a high degree of presence in VR.

Caching of frequently accessed data in ad hoc networks is a potential technique that can improve the data access performance and availability. Cooperative caching, which allows the sharing and coordination of cached data among clients, can further explore the potential of the caching techniques. In this paper, we propose a novel scheme, called zone cooperative (ZC) for caching in mobile ad hoc networks. In ZC scheme, one-hop neighbors of a mobile client form a cooperative cache zone since the cost for communication with them is low both in terms of energy consumption and message exchange. For a data miss in the local cache, each client first searches the data in its zone before forwarding the request to the next client that lies along routing path towards server. As a part of cache management, cache admission control and replacement policy are developed to improve the data accessibility and reduce the local cache miss ratio. Simulation experiments show that the ZC caching mechanism achieves significant improvements in terms of cache hit ratio and average query latency in comparison with other caching strategies

The integrated satellite-terrestrial network is promising to provide global broadband communication service. However, the long propagation delay of satellite-terrestrial links will lead to high communication delay when users access the Internet via satellites. In this paper, we investigate the cooperative multilayer edge caching in the integrated satellite-terrestrial network to reduce the communication delay, in which the base station cache, the satellite cache, and the gateway cache cooperatively provide content service for ground users. We first propose the three-layer cooperative caching model of the network, based on which we analyze the content retrieving process and derive the cache hit probability for different caching locations. Considering limited cache sizes, we formulate the content placement problem to minimize the average content retrieving delay of users. Then, two caching strategies, the non-cooperative caching strategy and the cooperative caching strategy, are proposed with exhaustive theoretical analysis. By introducing the concept of delay reduction gains, the optimal caching strategies are obtained based on the proposed iterative algorithms. Finally, numerical results are presented to demonstrate the performance of the proposed cooperative caching architecture and the caching strategies.

We present GLTScene, a novel data‐driven method for high‐quality furniture layout synthesis with general room boundaries as conditions. This task is challenging since the existing indoor scene datasets do not cover the variety of general room boundaries. We incorporate the interior design principles with learning techniques and adopt a global‐to‐local strategy for this task. Globally, we learn the placement of furniture objects from the datasets without considering their alignment. Locally, we learn the alignment of furniture objects relative to their nearest walls, according to the alignment principle in interior design. The global placement and local alignment of furniture objects are achieved by two transformers respectively. We compare our method with several baselines in the task of furniture layout synthesis with general room boundaries as conditions. Our method outperforms these baselines both quantitatively and qualitatively. We also demonstrate that our method can achieve other conditional layout synthesis tasks, including object‐level conditional generation and attribute‐level conditional generation. The code is publicly available at https://github.com/WWalter-Lee/GLTScene.

This paper presents CMP Cooperative Caching, a unified framework to manage a CMP's aggregate on-chip cache resources. Cooperative caching combines the strengths of private and shared cache organizations by forming an aggregate "shared" cache through cooperation among private caches. Locally active data are attracted to the private caches by their accessing processors to reduce remote on-chip references, while globally active data are cooperatively identified and kept in the aggregate cache to reduce off-chip accesses. Examples of cooperation include cache-to-cache transfers of clean data, replication-aware data replacement, and global replacement of inactive data. These policies can be implemented by modifying an existing cache replacement policy and cache coherence protocol, or by the new implementation of a directory-based protocol presented in this paper. Our evaluation using full-system simulation shows that cooperative caching achieves an off-chip miss rate similar to that of a shared cache, and a local cache hit rate similar to that of using private caches. Cooperative caching performs robustly over a range of system/cache sizes and memory latencies. For an 8-core CMP with 1MB L2 cache per core, the best cooperative caching scheme improves the performance of multithreaded commercial workloads by 5-11% compared with a shared cache and 4-38% compared with private caches. For a 4-core CMP running multiprogrammed SPEC2000 workloads, cooperative caching is on average 11% and 6% faster than shared and private cache organizations, respectively.

As most of the supply chains in practice are decentralized, a key issue in supply chain management is to study the coordination of supply chain. The interactions among players are typically of two types, viz. vertical competition and horizontal competition. We use a game theoretic framework to analyze a two-stage supply chain. The supply chain is modeled as a Stackelberg game with a manufacturer as leader and multiple retailers as followers. We consider multiple retailers competing on quantities (Cournot competition) and study its impact on manufacturer's equilibrium decisions and profits. We show that the wholesale price contract does not achieve Nash equilibrium when retailers offer uniform pricing. On the other hand, in case of perfect price discrimination we prove the existence of unique Nash equilibrium in both deterministic and stochastic demand set-up. Under stochastic demand set-up, we consider information asymmetry between the manufacturer and retailers with respect to demand signal. Furthermore, we show the retailers' selling effort boost up the sales. As variance of sales can hurt the manufacturer's production decisions badly, we conjecture that the manufacturer can offer a retailer incentive contract which can manipulate retailers' equilibrium decision.

As a promising architectural design for future Internet, named data networking (NDN) relies on in-network caching to efficiently deliver name-based content. However, the in-network caching is vulnerable to cache pollution attacks (CPA), which can reduce cache hits by violating cache locality and significantly degrade the overall performance of NDN. To defend against CPA attacks, the most effective way is to first detect the attacks and then throttle them. Since the CPA attack itself has already imposed a huge burden on victims, to avoid exhausting the remaining resources on the victims for detection purpose, we expect a lightweight detection solution. We thus propose ELDA, an Efficient and Lightweight Detection scheme against cache pollution Attacks, in which we design a Lightweight Flajolet-Martin (LFM) sketch to monitor the interest traffic. Our analysis and simulations demonstrate that, by consuming a few computation and memory resources, ELDA can effectively and efficiently detect CPA attacks.

We consider a network model where small base stations (SBSs) have caching capabilities as a means to alleviate the backhaul load and satisfy users’ demand. The SBSs are stochastically distributed over the plane according to a Poisson point process (PPP) and serve their users either (i) by bringing the content from the Internet through a finite rate backhaul or (ii) by serving them from the local caches. We derive closed-form expressions for the outage probability and the average delivery rate as a function of the signal-to-interference-plus-noise ratio (SINR), SBS density, target file bitrate, storage size, file length, and file popularity. We then analyze the impact of key operating parameters on the system performance. It is shown that a certain outage probability can be achieved either by increasing the number of base stations or the total storage size. Our results and analysis provide key insights into the deployment of cache-enabled small cell networks (SCNs), which are seen as a promising solution for future heterogeneous cellular networks.

Fog computing is today considered a promising candidate to improve the user experience in dynamic on-demand computing services. However, its ubiquitous application would require support for this service in wireless multi-hop mesh systems, where the use of conventional IP-based solutions is challenging. As a complementary solution, in this paper, we consider a Named-Data Networking (NDN) approach to enable fog computing services in autonomous dynamic mesh formations. In particular, we jointly implement two critical mechanisms required to extend the NDN-based fog computing architecture to wireless mesh systems. These are (i) dynamic face management systems and (ii) a learning-based route discovery strategy. The former makes it possible to solve NDN issues related to an inability to operate over a broadcast medium. Also, it improves the data-link layer reliability by enabling unicast communications between mesh nodes. The learning-based forwarding strategy, on the other hand, efficiently reduces the amount of overhead needed to find routes in the dynamically changing mesh networks. Our numerical results show that, for static wireless meshes, our proposal makes it possible to fully benefit from the computing resources sporadically available up to several hops away from the consumer. Additionally, we investigate the impacts of various traffic types and NDN caching capabilities, revealing that the latter result in much better system performance while the popularity of the compute service contributes to additional performance gains.