Blockchain-based audit trails provide a consensus-driven and tamper-proof trail of system events that are helpful in creating provenance in enterprise solutions. However, taking into account the transaction bulk generated by these applications and the throughput limitations of existing blockchains, a single ledger for record keeping can be inefficient and costly. To that end, we see an imperative need for a new blockchain design that is capable of addressing current challenges, without compromising security and provenance. Hence, we propose BlockTrail, a scalable and efficient blockchain solution for auditing applications. BlockTrail fragments the legacy blockchain systems into layers of co-dependent hierarchies, thereby reducing the time and space complexity, and increasing the throughput. BlockTrail is prototyped on "Practical Byzantine Fault Tolerance" (PBFT) protocol with a custom-built blockchain. Experiments with BlockTrail show that compared to the conventional schemes, BlockTrail is more efficient, and has less storage footprint.

The popularity of IoT has raised grave security and privacy concerns. There is a misconception that security and privacy issues of IoT systems are caused by the hardware and its cost. In this paper, we will explore the use of microcontrollers (MCUs) and crypto modules in IoT applications and demonstrate that hardware and cost may not be the bottleneck of IoT security and privacy in various application domains. We discuss how to implement hardware security, system/firmware security, network security, and data security with the low-cost Espressif's ESP32, TI's CC3220 and Microchip's cryptographic co-processor ATECC608A. We perform extensive experiments to validate the performance of cryptographic and networking operations of IoT devices based on those and other MCUs and crypto modules. We are the first to perform a comprehensive measurement and comparison of cryptographic and networking performance of these modern IoT MCUs and modules.

The cloud data centers are going through an unprecedented growth from past few years. In an era of real-time video streaming, on-demand gaming, door-step e-commerce services, and highly inter-connected social networks, cost-effective service models, adaptive resources provisioning and upfront applications availability contribute significantly towards such a stellar growth. However, there are many challenges that must be addressed in a systematic manner to meet the requirements of increasingly demanding current and upcoming applications of the cloud computing paradigm. Optimum resources management, instant response time, interoperability among a diverse set of emerging technologies and innovative applications are a few of these challenges. On the other hand, the recent trend in softwarization of networks, particularly enabled by network function virtualization (NFV) and software-defined networking (SDN) principles, provides immense opportunities to better utilize the network resources by programmable abstractions with an efficient control and management techniques. Furthermore, machine learning based solutions are gaining prominence in resource optimization problems and autonomous systems. Therefore, in this paper, we strive to connect the dots by state-of-the-art methodologies in networking and machine learning domains and utilize these developments to grapple with the challenges of the cloud-based systems. We propose DeepSDN, an SDN-based solution that harnesses existing machine learning techniques to move a step closer towards self-driving networks. The comparative results obtained from an experimental testbed corroborates effectiveness of our approach and suggest a way forward towards autonomous network management.

Audit logs serve as a critical component in the enterprise business systems that are used for auditing, storing, and tracking changes made to the data. However, audit logs are vulnerable to a series of attacks, which enable adversaries to tamper data and corresponding audit logs. In this paper, we present BlockAudit: a scalable and tamper-proof system that leverages the design properties of audit logs and security guarantees of blockchains to enable secure and trustworthy audit logs. Towards that, we construct the design schema of BlockAudit, and outline its operational procedures. We implement our design on Hyperledger and evaluate its performance in terms of latency, network size, and payload size. Our results show that conventional audit logs can seamlessly transition into BlockAudit to achieve higher security, integrity, and fault tolerance.

One of the challenges faced by data-intensive computing is the problem of stragglers, which can significantly increase the job completion time. Various proactive and reactive straggler mitigation techniques have been developed to address the problem. The straggler identification scheme is a crucial part of the straggler mitigation techniques, as only when stragglers are detected not only correctly but also early enough, the improvement in job completion time can make a real difference. Although the classical standard deviation method is a widely adopted straggler identification scheme, it is not an ideal solution due to certain inherent limitations. In this paper, we present Tukey's method, another statistical method for outlier detection, which is more suitable for the identification of stragglers for two reasons. First, it is robust to extreme observations from stragglers. Second, it can identify stragglers and, more importantly, start speculative execution earlier than the standard deviation method. Our extensive simulation results confirm that Tukey's method can remarkably outperform the standard deviation method.

Load balance is a crucial issue for data-intensive computing on cloud platforms, because a load balanced cluster can significantly improve the completion time of data-intensive jobs. In this paper, we present an improved replica placement policy for Hadoop Distributed File System (HDFS), which is specifically designed for heterogeneous clusters. The HDFS replica placement policy cannot generate balanced replica assignment, and hence has to rely on a load balance utility to balance the load among cluster nodes. In contrast, our proposed policy can generate perfectly even replica assignment, and also achieve load balance among cluster nodes in any heterogeneous or homogeneous environments without the running of the load balance utility.

Control plane scalability is one of the major concerns in Software Defined Networking (SDN) deployment. Although the centralization of the control plane by decoupling it from the data plane facilitates ease of network management, however, it introduces new challenges. One of these challenges is to maintain performance, consistency, and scalability while minimizing the corresponding overheads. In this paper, we propose an architecture that allows the control plane to evolve at a hyper-scale level as well as address important performance and reliability issues. A hierarchical control plane architecture with peer-to-peer communication among logically distributed controllers is designed with the goal of achieving optimum performance and consistency gains while mitigating overheads. A root controller is deployed at the top layer of the hierarchy to maintain global network view. The proposed model is helpful in improving network robustness against failures and supporting a desired level of reliability. To evaluate our model, we developed a realistic emulation platform using ONOS, FlowVisor, Mininet, and Open vSwitch. The proposed architecture is compared with earlier solutions and experimental results are presented to demonstrate the effectiveness of the proposed model.

Cloud based datacenters will be most suitable candidates for future software defined networking. The QoS requirements for shared data centers, hosting diverse applications could be successfully achieved through SDN architecture. This paper provides an extension of our previously proposed scheme QAMO that was aimed at achieving tangible QoS in datacenters through controlling bandwidth reservation in Multipath TCP and OBS layer while maintaining throughput efficiency. However, QAMO was designed for traditional networks and did not have the capability to adapt to current network status as expected from future software defined networks. The paper presents an enhanced algorithm called QAMO-SDN that introduces a controller layer in previously proposed architecture and achieves adaptive QoS differentiation based on current network feedback. QAMO-SDN inherits the architecture of QAMO, using Multipath TCP over OBS networks. We evaluate the performance of QAMO-SDN under different network loads and topologies using realistic data center traffic models and present the results of our detailed simulation tests.

Session hijacking attacks of social network websites are one of the commonly experienced cyber threats in today’s Internet especially with the unprecedented proliferation of wireless networks and mobile applications. To address this problem, we propose a cache supported hybrid two-dimensional one-wa

Wireless sensor networks are collections of tiny sensors working collaboratively to gather information. Two major shortcomings of today's WSNs are scarce energy and memory resources making it difficult to devise robust and effective security measures. Keyed-hash chains have been proposed as an efficient solution. Nevertheless, the basic configuration of one-way hash chains, where a shared key is hashed with no limits on the length of the chain causes an unwanted computational overhead and a potential security breach. In this paper, we propose a mini one-way hash chain protocol for user and broadcast authentication that is as effective, while at the same time significantly reducing the computational overhead. We avoid the potential security breach by using easily computed one time authentication tokens to secure communication. We also demonstrate that our protocol is energy preserving, light and efficient. Our simulation and evaluation tests reflect its benefit over straightforwardly configured one-way hash chains.

The use of insecure cookies as a means to authenticate web transactions in collaborative and social media websites presents a hazard to users’ privacy. In this paper, we propose and evaluate a novel protocol for protecting transmitted cookies using two dimensional one-way hash chains. In the first d

Social networking websites have been increasingly popular in the recent years. The users create and maintain their social networks by themselves in these websites by establishing or removing the connections to friends and sites of interests. The smart phones not only create a high availability for social network applications, but also serve for all forms of digital communication such as voice or video calls, e-mails and texts, which are also the ways to form or maintain our social network. In this paper, we deal with the problem of automatically generating and organizing social networks by analyzing and assessing mobile phone usage and interaction data. We assign weights to the different types of interactions. The interactions among users are then evaluated based on these weight values for certain periods of time. We use these values to rank the friends of users by a sports ranking algorithm, which recognizes the changes in the collected data over time.

One-way hash chains are a popular cryptographic technique used in many security applications. In this paper, we present a two-tier one-way hash chain (TTOHC) protocol to secure cookie-based Internet transactions. The use of cookies as a cheaper alternative to secure Internet sessions is extremely dangerous since session cookies can be easily sniffed out resulting in session hijacking. By utilizing different cryptographic hash functions arranged in two tiers, our hierarchical TTOHC protocol gives significant performance improvement over previously proposed solutions for securing Internet cookies. A detailed Java testbed has been used to evaluate alternative configurations for the hierarchical scheme and investigate the optimal set up of the two tiers. Detailed performance results obtained from this testbed are presented and analyzed.

The proliferation of social and collaborative media has been accompanied by an increased level of cyber attacks on social networking and collaboration sites. One serious type of attack is session hijacking attacks which enable the attacker to impersonate the victim and take over his/her networking s

In most existing studies of optical burst switching networks, adaptive routing is based on deflection routing and/or feedback from the past intervals which often introduce excessive transmission delay and architectural complexity. Our proposed three novel adaptive routing schemes, however, consider the transient link congestion at the moment when the bursts arrive; moreover they can utilize the same offset times for the same node pairs implying zero additional transmission delay and simplicity. The proposed schemes also aim to address the intrinsic unfairness defect of existing popular signaling protocols by increasing the effective link utilization. In performance results, we proceed to show that with our routing techniques, the fairness among the bursts with different hop counts can be greatly improved; and our methods can decrease the burst loss probability significantly as well in large-scale optical burst switching WDM mesh networks.

We proposed earlier three novel adaptive routing schemes which aimed to improve both loss fairness among bursts with various hop counts and overall burst loss performance in optical burst switched networks. This paper presents a reduced load fixed point approximation analysis to evaluate burst loss probabilities for symmetrical ring networks operating under the previously proposed routing. The analysis is based on a two-dimensional Markov chain model and its accuracy is validated by simulation.

In this paper, we present analytical models for the average packet service time, end-to-end delay and system capacity for the p-persistent 802.11 MAC protocol. Models for both the "no retry limit" and "finite retry limit" cases are presented. The analytical models are validated by extensive simulation results. The advantage of the P-persistent 802.11 MAC protocol over the legacy 802.11 MAC protocol in providing throughput differentiation is illustrated.

One of the major applications of sensor networks in near future will be in the area of biomedical research. Implanted biosensor nodes are already being used for various medical applications. These in-vivo sensor networks collect different biometric data and communicate the data to the base station wirelessly. These sensor networks produce heat, as the nodes have to communicate among themselves wirelessly. The rise in temperature of the nodes due to communication should not be very high. A high temperature of the in-vivo nodes for a prolonged period might damage the surrounding tissues. In this paper, we propose a new routing algorithm that reduces the amount of heat produced in the network. In the simple form, the algorithm routes packets to the coolest neighbor without inducing routing loops. In the adaptive form, the algorithm uses mechanisms to adapt to topologies with low degree of connectivity and to switch to shortest path routing if a time threshold is exceeded. The proposed algorithm performs much better in terms of reducing the amount of heat produced, delay and power consumption compared to the shortest hop routing algorithm and a previously proposed thermal aware routing algorithm (TARA)

In this paper, we present a new analytical model that captures link dependencies in all-optical wavelength-division multiplexing (WDM) networks under uniform traffic and enables the estimation of connection-blocking probabilities more accurately than previously possible. The basic formula of the dependency between two links in this model reflects their degree of adjacency, the degree of connectivity of the nodes composing them, and their carried traffic. Our validation tests have shown that the analytical dependency model gives accurate results and successfully captures the main dependency characteristics observed in the simulation measurements. The usefulness of the model is illustrated by showing how to use it in enhancing a simulation-based algorithm that we recently proposed for the sparse placement of full wavelength converters in WDM networks. To analytically handle the presence of wavelength converters, a lightpath containing converters is divided into smaller subpaths, such that each subpath is a wavelength-continuous path, and the nodes shared between these subpaths are full wavelength-conversion-capable. The blocking probability of the entire path is obtained by computing the probabilities in the individual subpaths. We validate the analytically-based sparse placement algorithm by comparing it with its simulation-based counterpart using a number of network topologies.

In this paper, we present a Web caching and pre-fetching scheme based on the concept of object flows. The scheme maintains a weighted directed flow graph whose vertices represent the objects accessed by the users and whose edges represent the reference flow relationships between these objects. Each directed edge in the graph is weighted in proportion to the frequency of occurrence of the reference sequence represented by that edge. Edge weights are adaptively changed to capture new access patterns and a controllable threshold parameter is used to prune out weak flow relationships. The decision for object pre-fetching or pre-eviction utilizes routines to compute objects that are strongly flow-related to the accessed/evicted object. The performance of the scheme was compared with LRU using an NS.2 simulation model. Our simulation used real life Internet traces from NANLR and UCB. An Oracle database was used to store/update the reference histories and was connected to the NS simulator using OCI (Oracle Call Interface). Extensive simulation tests were conducted to compare the performance of the two algorithms in terms of hit ratio and cost of pre-fetching. The tests covered a wide range of parameter values, e.g., cache size, the threshold for strong flow relationship, and the maximum number of objects that can be pre-fetched at a time. Our simulation experiments showed that the flow-based scheme is easy to implement and achieves significant improvement over LRU.

Abstract In this paper, we develop and evaluate a hierarchical cellular architecture for totally mobile wireless networks (TMWNs). Extensive performance tests were conducted to evaluate the performance of a two‐tier system and compare its throughput, handoff blocking rate and new call success rate with those obtained by a one‐tier model. Our tests have shown that when the total number of channels is kept the same, the two‐tier system outperformed the one‐tier counterpart under all load conditions. Under the constraint of equal power consumption, the two‐tier system still achieved improvement over the one‐tier system, especially at light and medium load levels. The improvement of the two‐tier system over the one‐tier system was observed to diminish as the degree of randomness in the mobility model is reduced; scenarios where the one‐tier system outperforms the two‐tier system are given. Load balancing schemes based on the concept of reversible handoffs are introduced and their performance improvements are analyzed. Comparison results on the percentage of terminal coverage are presented. An analytical model to compute the new call and handoff blocking probabilities in TMWN is given and evaluated. The model extends the Markov chain approach previously used in hierarchical architectures with stationary base stations and uses a corrected derivation for the handoff blocking probability. Copyright © 2002 John Wiley & Sons, Ltd.

Hierarchical cellular networks with mobile routers were proposed in our previous work to migrate the hierarchical feature from stationary cellular networks into totally mobile wireless networks (see Cui, W. and Bassiouni, M.A., Proc. IEEE Int. Conf. on Third Generation Wireless Communications, 2000). We briefly cover the basic idea of the proposed architecture and its channel planning issues. Then we introduce a fault recovery protocol using backup routers. Simulation results are presented for evaluation purposes. In addition, the steady-state performance of the recovery protocol is analyzed with stochastic models.

We propose and evaluate new schemes for channel reservation motivated by the rapidly evolving technology of mobile positioning. The schemes, called predictive channel reservation (PCR), work by sending reservation requests to neighboring cells based on extrapolating the motion of mobile stations (MSs). A number of design enhancements are incorporated to minimize the effect of false reservations and to improve the throughput of the cellular system. These enhancements include: (1) reservation pooling; (2) queuing of reservation requests; (3) hybrid approach for integrating guard channels (GCs); and (4) using a threshold distance (TD) to control the timing of reservation requests. The design enhancements have produced a set of highly efficient schemes that achieve significant reduction in handoff blocking rates while only incurring remarkably small increases in the new call blocking rates. The PCR approach has also been used to solve the MINBLOCK optimization problem and has given significant improvement over the fractional guard channel (FGC) protocol. Detailed performance results of the different variations of the PCR scheme and comparisons with conventional channel reservation schemes are presented. An analytical Markov model for the hybrid predictive version of the scheme is developed and its applicability and numerical results are discussed.

The wide deployment of multimedia services in third generation wireless networks will require designs that can simultaneously reduce the blocking probability of handoff requests and decrease the handoff delay. In this paper, we discuss this type of designs for dynamic channel assignment and reassignment strategies in highway cellular networks with a small channel reuse distance. A threshold-based approach to reduce the number of messages exchanged among base stations is presented. The properties of the two real-time schemes covered in this paper are presented and the global balance equations for the first scheme is derived based on the cell decoupling approximation model.

Data filtering is a semantic-based technique that can significantly reduce the size of traffic in large-scale distributed interactive simulation (DIS) networks. If not carefully designed, filtering algorithms could lead to errors that may compromise the realism of the training exercise. The authors review the problem of filtering errors and present an approach suitable for preventing filtering errors.

In this article, we discuss the problem of enhancing the speed of Huffman decoding. One viable solution to this problem is the multibit scheme which uses the concept of k-bit trees to decode up to k bits at a time. A linear-time optimal solution for the mapping of 2-bit trees into memory is presented. The optimal solution is derived by formulating the memory mapping problem as a binary string mapping problem and observing that at most four different 4-bit patterns can occur within any 2-bit Huffman tree. In addition to reducing the processing time of decoding, the optimal scheme is storage efficient, does not require changes to the encoding process, and is suitable for hardware implementations. © 1995 John Wiley & Sons, Inc.

Standard Boolean and comparison operators are based on the True/False logic and are therefore not quite suitable for handling events/activities defined over periods of time. In this paper, we generalize the Boolean and comparison operators by allowing their input and output operands to be sets of time intervals. Efficient algorithms for the implementation of the extended operators are presented. The generalized operators can be used to enhance the user interface of time-oriented query languages and allow application users to express their time-related qualifications more succinctly and elegantly. The paper is concluded by examples showing the flexibility of the extended operators in expressing time-related requirements.

Dynamic congestion control strategies for supporting real-time multimedia applications in high-speed integrated asynchronous transfer mode (ATM) networks are considered. Several traffic types including voice, video and multimedia data are expected to co-exist in future broadband integrated services digital network (B-ISDN). They are classified according to their distinct needs based on bandwidth, delay or delay-jitter and cell-loss rates. The significant merits that congestion control schemes in high-speed networks should have, in sharp contrast to conventional flow-based schemes, are identified. It is observed that a multilevel congestion control approach is essential to provide quality of service (QOS) guarantees in multimedia networks. A new service mechanism known as fast burst sharing is discussed. It has the ability to transport bursts of certain sessions without fragmenting them, thereby preserving integrity of user talkspurt, often desired in a bursty traffic environment. A number of dynamic cell marking schemes are proposed as effective traffic-compliance testing algorithms.

A hierarchical broadcasting technique in very large arbitrary networks which achieves near optimal cost and time measures is proposed. The network is partitioned into hierarchical clusters which are interconnected with each other through gate nodes. The approach is based on using a fully distributed minimum spanning tree algorithm within the gate-node network while achieving time optimal broadcasting within the local clusters by formulating the problem as finding maximum matching in constrained bipartite graphs. Extensive simulation results are presented which show the tradeoff between the two conflicting performance criteria.>

We present new memory based architectures for the design of special purpose hardware for real-time compression and decompression of data. The architecture is based on a novel idea of mapping the decode/encode tree of any binary code on to a memory device that corresponds to simultaneous decoding/encoding of multiple bits. The hardware is programmable, adaptable and yields a high compression rate. Using 1 μm CMOS process technology, this could easily lead to over 100 Mbits/sec compression rate for the JPEG baseline compression scheme.

A new compression scheme is described that aims at improving the fidelity of reconstructed images through the parallel application of both lossless and lossy compression techniques. The purpose of the scheme is to obtain compression ratios higher than those obtained by lossless compression schemes and at the same time produce reconstructed images with better fidelity than those normally obtained with lossy techniques. Tests so far have shown that the integrated lossless/lossy (IL/L) compression scheme consistently improves the fidelity of reconstructed images compared to well-known image compression algorithms. Specifically, the new scheme gives better fidelity (20% to 55% reduction in mean square error) than the DCT algorithm under equal compression ratios.>

An approach and a mechanism for the definition and maintenance of views within temporal database environments are outlined. The view definition language makes use of extending Boolean and comparison operators by allowing their operands to be sets of intervals. A mechanism for maintaining temporal views is presented, and it is shown that temporal database views have far fewer maintenance problems than their standard relational counterparts.>

Schemes that can be used to compress image data as well as nonimage database files efficiently are presented. An initial transformation (coding) step is applied to image data in order to induce desirable properties that enable the efficient application of popular lossless compression schemes that are normally well suited to text and database files. Further improvement is obtained by a scheme that utilizes the correlation properties of the transformed image files.>

A new class of VLSI architectures for data transformation of tree-based codes is proposed. The focus is on transformation functions used for data compression and decompression. The encoding algorithm is based on a pipeline architecture and can generate the code bits in parallel. The algorithms use the principle of propagation of a token in a reverse binary tree constructed from the original codes. The design approaches are applicable to any binary codes, although static Huffman code is used as an illustration. A new hardware algorithm for generating adaptive Huffman codes is proposed and a VLSI architecture for implementing the algorithm is described. The high speed of the new algorithms ensures that data transformation is done as data are being transferred from/to high-speed I/O communication devices.>

Methods are discussed to enhance the efficiency and speed of data compression techniques in DBMS (database management systems). Arithmetic coding utilizes the skewness of character distribution by assigning larger intervals (code ranges) to characters having higher probabilities of occurrence. A scheme is presented which effectively increases the code ranges of individual characters by splitting the interval assignment into different groups. This decreases the rate of interval narrowing and hence improves the compression efficiency. Hardware assistance for arithmetic and tree-based coding is also discussed and high-speed VLSI algorithms for data compression are presented. The proposed algorithms give rates that are an order of magnitude faster than currently attainable encoding speeds.>

An approach to the handling of time in query languages of historical database management systems is outlined. The approach is based on extending Boolean and comparison operators by allowing their operands to be sets of time intervals. The proposed temporal logic is shown to satisfy the properties of the normal Boolean logic. A relational-calculus query language using the extended logic is proposed. A new syntax for retrieval statements is defined in order to separate the process of selecting entities (tuples) and the process of selecting required values of temporal attributes from the chosen entities. The extension presented in this paper offer a good degree of flexibility in expressing different temporal requirements.>

A compression algorithm that is tailored to utilize the enormous speed and memory size of supercomputers is presented. It utilizes an enhanced arithmetic coding scheme. Efficient VLSI designs for the enhanced scheme are given. The designs are suitable for inclusion in disk and communication controllers of supercomputers and would provide a practical way to increase thresholds of data transfer rates and communication bandwidth effectively.>

The authors consider that, in spite of their advantage in removing the overhead of lock maintenance and deadlock handling, optimistic concurrency control methods have been applied less in practice than locking schemes. Two complementary approaches are introduced that may help render the optimistic approach practically viable. For the high-level approach, integration schemes can be utilized so that the database management system is provided with a variety of synchronization methods each of which can be applied to the appropriate class of transactions. The low-level approach seeks to increase the concurrency of the original optimistic method and improve its performance. The author examines the low-level approach in depth, and presents algorithms that aim at reducing back-ups and improve throughput. Both the single-site and distributed networks are considered. Optimistic schemes using time-stamps for fully duplicated and partially duplicated database networks are presented, with emphasis on performance enhancement and on reducing the overall cost of implementation.>

In this paper, we present a functional description of a VLSI chip aimed at reducing the cost of data transmission and data access within information processing machines and distributed information systems. The chip maps standard character codes (e.g., ASCII) into more efficient codes (e.g., Huffman's codes) using a tree module of basic cells. In bit-serial communication controllers, for example, the parallel-to-serial transformation unit can be simply replaced by the proposed chip. The VLSI design can provide speeds that far exceed current and projected peak transfer rates of high-speed disks and communication controllers.

Two models for the storage and placement of files on disk storage devices are presented. Both models assume knowledge of certain aspects of the I/O behaviour in the system. The models are based on the system viewpoint and the program viewpoint, respectively, of I/O measurements and they both use the seek delay time as the performance measure to be optimised. The underlying analytic problem in each case is NP-hard. Efficient heuristic algorithms are presented and numerically tested.

Scientific and statistical database systems heavily depend on data compression techniques to make possible the management and storage of their large databases. The efficiency of data compression methods has a signficant impact on the overall performance of these systems. The purpose of this paper is to show the importance of data compression to scientific/statistical databases, to discuss the pros and cons of data compression, and to survey data compression techniques relevant to scientific/statistical databases. The emphasis is on the basic idea, motivation, and tradeoffs of each approach. Both software and hardware methods are covered. The paper is concluded by a discussion of several points of research that seem worthy of further investigation.

An efficient compression technique that is particularly attractive for the storage of large commercial files and the transfer of such files within a distributed communication network is outlined. The technique, constructed as a two-level hierarchy of Huffman-type binary trees, is a reversible semantic-independent variable-length character encoding method that makes use of the group locality of character reference behavior and the variable frequency of occurrence of various characters within the different subgroupings of the character set The compression efficiency and cost of the Multi-Group scheme has been found to be superior to the well-known Huffman encoding algorithm.