Edge-based distributed deep learning (DDL) is increasingly important for embedded AI scenarios where data resides within a trusted domain (e.g., a hospital or campus network) and compute is pushed to edge nodes for cost or performance reasons. Deploying efficient DDL systems in such settings requires evaluating a large space of training configurations, each defined by a combination of cluster topology and synchronization strategy. Exploring this space through real training is prohibitively slow, and existing single-machine simulators are inaccurate and inefficient, as they require full training runs to estimate timing, overlook communication effects, and suffer timing distortions due to host-level scheduling.

In this paper, we investigate the problem of achieving efficient memory resource management for unikernel-based virtual machines (uVMs), where unikernels are running as the operating systems of the VMs. Through extensive experiments, we first demonstrate that existing VM memory management mechanisms are unsuitable for uVMs. Then, we propose MEUNIK, a system that aims to achieve high memory management efficiency or uVMs. The four key mechanisms of MEUNIK are designed to address the problems we find in existing solutions. We have implemented a prototype MEUNIK system based on the Xen hypervisor and performed extensive evaluation experiments on three groups of uVMs, which are constructed based on a variety of programs and applications, including fifteen small benchmark programs, four complex server applications, and two network-operation-heavy programs. The evaluation results show that our system achieves the design goals with minimal overhead.

We identify and survey five mechanisms through which web content interacts with mobile apps. While useful, these web-app interaction mechanisms cause various notable security vulnerabilities on mobile apps or web content. The root cause is lack of proper access control mechanisms for web-app interactions on mobile OSes. Existing solutions usually adopt either an origin-centric design or a code-centric design, and suffer from one or several of the following limitations: coarse protection granularity, poor flexibility in terms of access control policy establishment, and incompatibility with existing apps/OSes due to the need of modifying the apps and/or the underlying OS. More importantly, none of the existing works can organically deal with all the five web-app interaction mechanisms. In this paper, we propose ReACt, a novel Resource-centric Access Control design that can coherently work with all the web-app interaction mechanisms while addressing the above-mentioned limitations. We have implemented a prototype system on Android, and performed extensive evaluation on it. The evaluation results show that our system works well with existing commercial off-the-shelf Android apps and different versions of Android OS, and it can achieve the design goals with small overhead.

We show that many popular mobile services suffer from excessive network bandwidth consumption. The root cause is that the existing mobile/cloud communication interfaces are designed and optimized for service providers rather than end user devices. Solving the problem is challenging, because of the conflicted interests of service providers and mobile devices. We propose Edge-hosted Personal Service (EPS), with which device-oriented solutions can be easily deployed without affecting service providers. EPS also enjoys other notable advantages, including enabling new mobile services, reducing loads on the cloud, and benefiting delay-sensitive applications. We demonstrate the usefulness of EPS by designing ETA (Edge-based web Traffic Adaptation), an effective solution for the excessive bandwidth consumption problem, and deploy ETA with a prototype EPS system. By exploring lightweight virtualization techniques, our EPS prototype system is highly scalable in terms of concurrent EPS instances, and secure in terms of resource isolation. The real-world evaluation shows that our ETA EPS can effectively reduce bandwidth for mobile devices with small overheads.

Cloud storage usages are becoming increasingly popular on mobile devices. Through an extensive motivation study, we find that cloud storage accesses from mobile apps suffer from several notable problems that undermine usage experiences. The root cause is that the way of cloud storage providers deploying their services onto mobile devices relies on app developers for the correct and appropriate implementations and lacks the ability of monitoring and servicing client-side cloud storage accesses. We propose StoArranger, a practical system framework that solves the problems by coordinating, rearranging, and transforming cloud storage communications on mobile devices. We have prototyped the proposed system using two different implementation approaches. We discuss our experiences of the implementations in the paper. The real-app evaluation experiments show that StoArranger can significantly improve mobile cloud storage access efficiency with little overheads.

CPU is one of the most significant sources of power consumption on smartphones. Power modeling is a key technique and important tool for power estimation and management, both of which are critical for providing good QoS for smartphones. However, we find that existing CPU power models for smartphones are ill-suited for modern multicore CPUs: they can give high estimation errors (up to 34%) and high estimation accuracy variation (more than 30%) for different types of workloads on mainstream multicore smartphones. The cause is that the existing approaches do not appropriately consider the effects of CPU idle power states on smartphones CPU power modeling. Based on our extensive measurement experiments, we develop a new CPU power modeling approach that carefully considers the effects of CPU idle power states. We present the detailed design of our power modeling approach, and a prototype CPU power estimation system on commercial multicore smartphones. Evaluation results show that our approach consistently achieves higher power estimation accuracy and stability for various benchmarks programs and real apps than the existing approaches.

Being powered by battery and the reliance on expensive cellular data while users are on the move are holding smart mobile devices from being widely used in those long-lasted, computation-intensive, or highly network-reliant usage scenarios. In the meantime, more and more mobile workloads and optimizations now rely on the cloud, such as mobile cloud offloading, cloud-based mobile web optimization, and cloud-based network traffic redundancy reduction. However, it is hard to perform large-scale and personalized support or optimization for mobile workloads without significant computation resource increase on the cloud, as well as higher bandwidth requirement on the networks. We target solving the above two problems by enabling edge-hosed personalized services (EPS for short). The idea of EPS is twofold. One is enabling developer-customizable and power-and-traffic-efficient network communication on the "last-hop" communication between mobile devices and the network edge. The other is distributing cloud services for mobile workloads to network edge, so that they can be done on a personalized basis, while enjoying much lower communication latency to/from mobile devices.

Smartphone device inputs, such as inputs from touchscreen, sensors, and GPS, carry sensitive user information, but are vulnerable to passive and active attacks. We present our ongoing design and implementation of SDIF, a Secure Device Input Framework for smartphones. The core components of SDIF are a small and dedicated bare-metal hypervisor built using ARM hardware virtualization support and a user-space sandbox framework. They collectively ensures SDIF's support for unmodified OSes and apps. SDIF guarantees the secure device inputs from hardware to target applications with two key designs. The first is a design of a secure path for input data to be delivered from input device drivers to target applications. The second is a design of trusted device data reading, which ensures input data generated by hardware is faithfully put to device driver memory.

In recent years mobile smart devices such as tablets and smartphones have exploded in popularity. We are now in a world of ubiquitous smart devices that people rely on daily and carry everywhere. This is a fundamental shift for computing in two ways. Firstly, users increasingly place unprecedented amounts of sensitive information on these devices, which paints a precarious picture. Secondly, these devices commonly carry many physical world interfaces. In this paper, we propose information leakage malware, specifically designed for mobile devices, which uses covert channels over physical "real-world" media, such as sound or light. This malware is stealthy; able to circumvent current, and even state-of-the-art defenses to enable attacks including privilege escalation, and information leakage. We go on to present a defense mechanism, which balances security with usability to stop these attacks.

Mobile peer-to-peer applications require devices to network themselves on-the-fly to communicate directly with each another. This paper presents mQual, a framework to help create such networks that meet different application requirements, and is able to adjust the network to ensure that these requirements are met in dynamic environments. Our prototype mQual extends the current WiFi-Direct in Android, and the experimental results suggests that mobile apps built using mQual outperform those built using WiFi-Direct.

Mobile devices are now ubiquitous in the modern world. In this paper, we propose a novel and practical mobile-cloud platform for smart mobile devices. Our platform allows users to run the entire mobile device operating system and arbitrary applications on a cloud-based virtual machine. It has two design fundamentals. First, applications can freely migrate between the user's mobile device and a backend cloud server. We design a file system extension to enable this feature, so users can freely choose to run their applications either in the cloud (for high security guarantees), or on their local mobile device (for better user experience). Second, in order to protect user data on the smart mobile device, we leverage hardware virtualization technology, which isolates the data from the local mobile device operating system. We have implemented a prototype of our platform using off-the-shelf hardware, and performed an extensive evaluation of it. We show that our platform is efficient, practical, and secure.

We present HoWiES, a system that saves energy consumed by WiFi interfaces in mobile devices with the assistance of ZigBee radios. The core component of HoWiES is a WiFi-ZigBee message delivery scheme that enables WiFi radios to convey different messages to ZigBee radios in mobile devices. Based on the WiFi-ZigBee message delivery scheme, we design three protocols that target three WiFi energy saving opportunities in scanning, standby and wakeup respectively. We have implemented the HoWiES system with two mobile devices platforms and two AP platforms. Our real-world experimental evaluation shows that our system can convey thousands of different messages from WiFi radios to ZigBee radios with an accuracy over 98 percent, and our energy saving protocols, while maintaining the comparable wakeup delay to that of the standard 802.11 power save mode, save 88 and 85 percent of energy consumed in scanning state and standby state respectively.

Cognitive radio (CR) networks have an ample but dynamic amount of spectrum for communications. Communication rendezvous in CR networks is the process of establishing a control channel between radios before they can communicate. Designing a communication rendezvous protocol that can take advantage of all the available spectrum at the same time is of great importance, because it alleviates load on control channels, and thus further reduces probability of collisions. In this paper, we present ETCH, efficient channel-hopping-based MAC-layer protocols for communication rendezvous in CR networks. Compared to the existing solutions, ETCH fully exploits spectrum diversity in communication rendezvous by allowing all the rendezvous channels to be utilized at the same time. We propose two protocols, SYNC-ETCH, which is a synchronous protocol assuming CR nodes can synchronize their channel hopping processes, and ASYNC-ETCH, which is an asynchronous protocol not relying on global clock synchronization. Our theoretical analysis and ns-2-based evaluation show that ETCH achieves better performances of time-to-rendezvous and throughput than the existing work.

Efficient web caching in mobile apps eliminates unnecessary network traffic, reduces web accessing latency, and improves smartphone battery life. However, recent research has indicated that current mobile apps suffer from poor implementations of web caching. In this work, we first conducted a comprehensive survey of over 1000 Android apps to identify how different types of mobile apps perform in web caching. Based on our analysis, we designed CacheKeeper, an OS web caching service transparent to mobile apps for smartphones. CacheKeeper can not only effectively reduce overhead caused by poor web caching of mobile apps, but also utilizes cross-app caching opportunities in smartphones. Furthermore, CacheKeeper is backward compatible, meaning that existing apps can take advantage of CacheKeeper without any modifications. We have implemented a prototype of CacheKeeper in Linux kernel. Evaluation on 10 top ranked Android apps shows that our CacheKeeper prototype can save 42% networks traffic with real user browsing behaviors and increase web accessing speed by 2x under real 3G settings. Experiments also show that our prototype incurs negligible overhead in most aspects on cache misses.

We propose HoWiES, a system that saves energy consumed by WiFi interfaces in mobile devices with the assistance of ZigBee radios. The core component of HoWiES is a WiFiZigBee message delivery scheme that enables WiFi radios to convey different messages to ZigBee radios in mobile devices. Based on the WiFi-ZigBee message delivery scheme, we design three protocols that target at three WiFi energy saving opportunities in scanning, standby and wakeup respectively. We have implemented the HoWiES system with two mobile devices platforms and two AP platforms. Our real-world experimental evaluation shows that our system can convey thousands of different messages from WiFi radios to ZigBee radios with an accuracy over 98%, and our energy saving protocols, while maintaining the comparable wakeup delay to that of the standard 802.11 power save mode, save 88% and 85% of energy consumed in scanning state and standby state respectively.

This paper considers vehicular rogue access points (APs) that rogue APs are set up in moving vehicles to mimic legitimate roadside APs to lure users to associate to them. Due to its mobility, a vehicular rogue AP is able to maintain a long connection with users. Thus, the adversary has more time to launch various attacks to steal users' private information. We propose a practical detection scheme based on the comparison of Receive Signal Strength (RSS) to prevent users from connecting to rogue APs. The basic idea of our solution is to force APs (both legitimate and fake) to report their GPS locations and transmission powers in beacons. Based on such information, users can validate whether the measured RSS matches the value estimated from the AP's location, transmission power, and its own GPS location. Furthermore, we consider the impact of path loss and shadowing and propose a method based on rate adaption to deal with advanced rogue APs. We implemented our detection technique on commercial off-the-shelf devices including wireless cards, antennas, and GPS modules to evaluate the efficacy of our scheme.

In a dynamic spectrum access (DSA) network, communication rendezvous is the first step for two secondary users to be able to communicate with each other. In this step, the pair of secondary users meet on the same channel, over which they negotiate on the communication parameters, to establish the communication link. This paper presents ETCH, Efficient Channel Hopping based MAC-layer protocols for communication rendezvous in DSA networks. We propose two protocols, SYNC-ETCH and ASYNC-ETCH. Both protocols achieve better time-to-rendezvous and throughput compared to previous work.