In this paper, we investigate the effects of violating the conditions of hash function uniformity and/or independence on the false positive probability of Bloom Filters (BF). To this end, we focus on hash functions of the H3 family with a partitioned memory organization for fast hardware implementations of BFs. We first introduce a dependence metric that quantifies hash function uniformity and independence. We then state and prove the necessary and sufficient conditions on the BF parameters for constructing uniform and independent hash functions. Finally, we derive an analytical expression for the exact false positive probability of a BF with hash functions that are not necessarily uniform or independent. We verify our expression with a hardware test bench and explore the effects of losing uniformity and independence through an experimental study that systematically sweeps different dependence metric values and numbers of hash functions. We demonstrate the effects of violating hash function uniformity and independence on the stated target false positive probability for selected previous works in the literature. As an important finding, we show that uniformity of individual hash functions is essential, whereas limited dependencies between hash functions can be tolerated without a negative effect on the false positive probability.

This paper proposes weak TDMA (WTDMA) as a new method for slotted medium access on the controller area network (CAN). Different from TDMA, WTDMA does not require guard times for the temporal isolation of different time slots. Instead, a certain degree of overlap of time slots can be tolerated due to the usage of Carrier Sense Multiple Access/Collision Resolution (CSMA/CR) and the non-preemptive message transmission on CAN. WTDMA can be realized in software with a moderate accuracy of the synchronized clocks of different CAN nodes and without any modifications to the CAN standard. The paper provides sufficient conditions for the correct operation of WTDMA and performs a comprehensive experimental evaluation. We show that bus loads above 90% can be achieved, whereby the message latencies are mostly determined by the message transmission times in the order of hundreds of microseconds. In addition, WTDMA is robust to disturbances such as bit errors. These properties are vital for contemporary mission-critical real-time vehicle applications.

Modern vehicles, that have to be considered as safety-critical cyber-physical systems, require highly accurateclock synchronization(CS) among their distributed computing devices. Since Controller Area Network (CAN) is the predominant in-vehicle communication bus, it is highly relevant to support CS for CAN. This article proposes an original CS method for distributed in-vehicle networks based on CAN with bothoffsetanddrift correction. While offset correction is performed based on timestamps in periodic reference messages (RMs), our new method benefits from the re-synchronization mechanism of the CAN bit timing to apply highly accurate drift correction. Our algorithm does not make any modifications to the CAN protocol but requires the measurement of the phase error from the CAN controller. We derive analytical bounds for the expected clock differences and further validate the practicability of the proposed method by comprehensive experiments. As the main result, our method achieves a clock accuracy below$2\;\mu$s independent of important parameters such as the bit rate, RM period, bus utilization and time-varying clock drifts.

Synchronizing clocks is crucial for distributed safety-critical in-vehicle applications. Since Controller Area Network (CAN) is the predominant in-vehicle communication bus, it is highly relevant to realize clock synchronization (CS) on CAN. This paper proposes a new software-based CS (SW-CS) algorithm based on the periodic transmission of reference messages (RMs) by a Master node and discrete-time feedback control. Different from existing algorithms, that only update clocks after receiving a RM, our algorithm corrects the clock drift between RMs. Measurements on a hardware setup show a decrease of the clock differences by more than one order of magnitude.

Network Slicing aims for creating isolated virtual networks called slices on the same physical infrastructure. The network resources are allocated to the slices according to their specific service requirements. This paper proposes Dynamic User Count Aware (DUCA) method for allocating radio access network resources to slices. Different than previous work DUCA accounts for the slice user count for deciding the amount of slice resources. To this end, DUCA increases the number of granted users in the network, increases the fairness of resource allocation towards slices with large number of users and small user bandwidth demands. The slices with small number of users and large bandwidth demands have better fitting resource allocation to their requests increasing the efficiency. We evaluate DUCA using simulation and compare its performance to relevant previous resource allocation frameworks. We observe that DUCA achieves the above benefits with comparatively better results.

Multifunction Vehicle Bus (MVB) is a highly robust real-time field bus for rail vehicles. On MVB, periodic process data and sporadic message data are transmitted in the form of telegrams. The main focus of this paper is the development of heuristic scheduling algorithms for periodic telegrams on MVB. We first propose two heuristics that find the best location of individual telegrams in the MVB schedule based on different criteria. Then, we introduce a new swap operation that allows improving MVB schedules by exchanging the locations of telegrams. A comprehensive evaluation based on a large number of test cases shows that the proposed heuristics are able to find feasible and close-to-optimal MVB schedules with practical computation times. In particular, our heuristics clearly outperform an existing ILP formulation, that cannot find optimal or even feasible schedules in cases with large telegram sets.

Cloud data center workloads have time- dependencies and are hence non-i.i.d (independent and identically distributed). In this paper, we propose a new model-based method for creating synthetic workload traces for cloud data centers that have similar time characteristics and cumulative distributions to those of the actual traces. We evaluate our method using the actual resource request traces of Azure collected in 2019 and the well-known Google cloud trace. Our method enables generating synthetic traces that can be used for a more realistic evaluation of cloud data centers.

Safety-critical in-vehicle applications require an accurate global time in order to coordinate their actions. Although Controller Area Network (CAN) is the most widely used in-vehicle communication bus, it does not support synchronized clocks. Hence, the realization of clock synchronization on CAN is highly relevant. This paper develops two new ideas for the clock synchronization on CAN. The first contribution is a new software-based algorithm based on periodic reference messages. Different from existing algorithms, the proposed algorithm enables the use of timestamps that are taken close to the reference message transmission hence improves the clock accuracy. The second contribution of the paper is an original method for correcting clocks between reference messages. In particular, the proposed algorithm updates local clocks based on the bit timing on the CAN bus. Results from simulation experiments show an improvement of about 60% in the clock precision.

In this paper we propose ACCLOUD-MAN, a novel resource manager for heterogeneous cloud data centers. In heterogeneous clouds a user request can be satisfied with more than one physical resource alternative. That is, the resource manager must decide which resource alternative will be chosen, along with the decision of the server the request will be assigned to. ACCLOUD-MAN's resource management objective is to reduce the power consumption of the cloud data center. To this end, the manager is modeled as an integer linear programming problem and is implemented in MATLAB, along with a cloud data center simulation platform. Simulation results show that the proposed ACCLOUD-MAN outperforms existing resource allocation methods such as Openstack.

OpenStack is a widely used management tool for cloud computing which is designed to work on servers and allocate standard computing resources such as CPU, memory or disk. The current trend for integrating different hardware accelerators such as FPGAs and GPUs in the cloud require managing these heterogeneous resources. In this paper, we propose a generalization for OpenStack Nova project which extends the relevant data structures to include these new resources. More importantly, we present a new lightweight Nova Compute module that we call Nova-G Compute that is suitable to work with different hardware platforms and can communicate with the rest of the OpenStack Projects. We implement a hypervisor-like software to enable Nova-G compute accessing the FPGA resources. We perform experimental evaluation of Nova-G using the known and used OpenStack benchmarking tool Rally. Our results show that Nova-G Compute works as desired without any reduced performance compared to standard Nova.

Many fields of computing such as Deep Packet Inspection (DPI) employ string matching modules (SMM) that search for a given set of positive strings in their input. An SMM is expected to produce correct outcomes while scanning the input data at high rates. Furthermore the string sets that are searched for are usually large and their sizes increase steadily. Bloom Filters (BFs) are hashing data structures which are fast but their false positive results require further processing. That is, their speed can be exploited for Standard Bloom Filter SMMs (SBFs) as long as the positive probability is low. Multiple BFs in parallel can further increase the throughput. In this paper, we propose the Double Bloom Filter SMM (DBF) which achieves a higher throughput than the SBF and maintains a high throughput even for large positive probabilities. The second Bloom Filter of DBF stores a small enough subset of the positive strings such that its false positive probability is approximately zero. We develop an analytical model of the DBF and show that the throughput advantage of DBF over SBF becomes more prominent if the positive probability and the fraction of matches in the second Bloom Filter increase. Accordingly, we propose a heuristic algorithm that stores the strings that are more frequently matched in the second Bloom Filter according to localities identified in the input. Our numerical results are obtained using realistic values from an FPGA implementation and are validated by SystemC simulations.

IP address lookup modules for backbone routers should store 100Ks of entries, find the longest prefix match (LPM) for each incoming packet at 10s of Gbps line speed and support thousands of lookup table (LUT) updates each second. It is desired that these updates are non-blocking, that is without disrupting the ongoing lookups. Furthermore, considering the increasing line rates and table sizes, the scalability of the design is very important. Ternary content-addressable memory (TCAM) architectures are widely deployed for hardware IP lookup. In this paper, we propose a novel TCAM architecture, S-DIRECT-Scalable and Dynamically REConfigurable TCAM, that is custom designed for hardware IP lookup. S-DIRECT consists of hierarchically combined TCAM cells with inherent priority encoders (PEs) to support LPM. Hence, its design is scalable without any need for a separate PE or a redesign for different table size. Furthermore, S-DIRECT can perform constant time, non-blocking updates in hardware provided that certain write capabilities are present in the TCAM entries. S-DIRECT architecture is both independent of the hardware platform and the implementation of the TCAM cells. We demonstrate the generality and viability of S-DIRECT by implementing it both with prefix/mask register and LUT-based TCAM cells on FPGA.

Multimedia applications over the Internet and Internet Protocol Television (IPTV) gain a lot of attention. IPTV has a number of service requirements such as; high bandwidth, scalability, bounds on delay, jitter and channel zapping time. IP multicast, IP Multimedia System (IMS) Protocol and peer-to-peer (P2P) approaches are proposed for implementing IPTV. However, IP multicast requires all the routers in the core network to possess multicast capability, IMS does not easily scale and P2P cannot efficiently utilize the network resources because of its completely distributed nature. To this end, we propose Cluster Based Application Layer Multicast (ALM) IPTV (CALMTV) which is a new application layer multicast protocol for live IPTV. CALMTV combines application layer multicast, scalable video coding and probing techniques to meet live IPTV requirements. We present the components and their relevant algorithms of CALMTV. The performance of CALMTV is demonstrated with ns2 simulations. Our experiment results compared with the published results of other ALM architectures, which can support IPTV, show that CALMTV improves end-to-end delay and zapping time with respect to these architectures without significant increase of jitter.

The services in the Next Generation Network (NGN) will be created on demand by the customers and will require end-to-end Quality of Service (QoS) for each flow. A very significant component for the end-to-end QoS support in the Internet is the packet schedulers in the routers. The complexity of the packet scheduling algorithms increases with the number of flows. As a solution, flow aggregation decreases the number of flows processed by the scheduler. The previous work in the literature proves that if the flow aggregator is fair, the end-to-end delay bounds of the aggregated flows are preserved and suggests limiting the service rate for the aggregate flow to achieve fairness in the expense of a lower utilization of the network resources. In this paper, we present a new method for flow aggregation, which relaxes this limit on the aggregate service rate, to increase the link utilization. We analytically show that our aggregation method is fair. Consequently, the end-to-end delay bounds in the network are preserved. In addition, we provide simulation results to demonstrate the decreased average delay of the aggregated flow.

The automation of today's large-scale industrial systems relies on the operation of distributed controller devices that perform local computations and exchange information via communication networks. The subject of this paper is the development of a family of shared-medium industrial communication protocols that support the transmission of real-time (RT) and nonreal-time (nRT) data among distributed controller devices. Different from existing protocols, we suggest to incorporate information that is available from the control application in the protocol definition. As a result, our protocols dynamically change the bandwidth allocation on the shared medium according to the instantaneous communication requirements while ensuring hard RT guarantees. Following the recent developments in industrial automation, our protocols can be realized as software layers on top of low-cost conventional Ethernet.

In this paper, we study the scheduling of periodic messages in the static segment of the FlexRay protocol. Our approach is based on two performance metrics. Similar to previous work, we seek to allocate a minimum number of frame IDs (FIDs) in the static segment. In addition, different from existing work, we want to minimize the message jitter, i.e., the deviation of the message transmission from the required periodicity. To this end, we first derive analytical expressions that quantify the FID allocation and the jitter, and then formulate a linear integer programming problem whose solution is the desired message schedule. An example illustrates our schedule optimization.

In this paper, we perform the schedulability analysis and schedule computation for sporadic real-time messages in the dynamic segment of the FlexRay protocol. We first formulate a linear integer programming problem that allows to determine if a given message schedule is feasible, i.e., the worst-case delay of each message is smaller than its deadline. Then, we develop a heuristic algorithm that enables the efficient computation of feasible schedules. Our results are illustrated by an experimental setup with three FlexRay nodes.