Executive Summary

A. Introduction

The National Information Infrastructure will usher in the Information Age for millions of Americans in all walks of life. This infrastructure will open up vast new markets in information services and will support large-scale applications in strategic areas such as health care, education, manufacturing, commerce, finance and government. End-user applications will range from basic data processing services, such as payroll and client billing, to advanced business and consumer services such as dependable routine and emergency access to medical data, programmable manufacturing processes that can quickly adapt to the changing demands of the marketplace, lifelong education and job training services.

The infrastructure will support these applications by providing:

1) Thousands of large information repositories that can be searched electronically for specified kinds of information (e.g., medical images and videos, electronic libraries, product information and business investment profiles).

2) Wide-bandwidth data networks and information appliances that provide every home and business in the nation with access to shared information repositories.

3) Advanced communication and information access services that can be used as building blocks for the development of affordable, interoperable, easy-to-use applications that have an appropriate level of security and dependability.

Because electronic delivery can provide tremendously greater accessibility and efficiency than traditional methods of delivering services, this information infrastructure, like the telephone system, has the potential to greatly improve the economic well-being, social welfare and quality of life for all Americans. The vision for the NII is that the economic welfare of the nation will be improved by the creation of entirely new markets, by services that allow businesses to compete successfully in international markets, by a better educated workforce, by increased efficiency and a reduced cost of providing government services, and by lower-cost, high-quality health care. The social welfare of the nation will be improved by access to government services and state-of-the-art educational opportunities (preschool through adult education and job training programs) through affordable information appliances that include the next-generation television sets. The quality of life will be improved by, for example, simpler access to government services, affordable access to quality health care, a stronger economy and vastly improved consumer services.

The information infrastructure envisioned above is many orders of magnitude more powerful than existing facilities such as the Internet. Its creation requires advancing into unknown territory, which in turn requires substantial research and development efforts aimed at 1) scaling up existing information technologies and services, 2) integrating these information technologies with advanced voice and video communication technologies, and 3) developing new technologies that address key issues such as advanced information storage and retrieval, ease of use, interoperability, dependability and manageability in this integrated, multimedia setting. These research and development efforts will take place in parallel with multi-supplier development of products that refine the results of the research and development efforts and collectively form the NII.

The realization of the NII will require long-term collaboration among industry, academia and government. A coordinated strategy with complementary investment is needed. The federal government has at least two significant roles to play in the development of the NII. The first is to devise and implement effective policies that enable the development of a coherent infrastructure while allowing competitive market forces to drive the creation of products and services that make up the infrastructure. The second is to foster and support a long-range research program to address many technical problems that are key to the success of the NII. The government has a particular responsibility to support research that has higher risk and/or longer horizons than the research that can be funded by industry. The government also has particular responsibility to support research for which great public benefit will be derived from the shared use of the research results. An appropriate federal research program, in which cooperating researchers from many institutions share experiences, approaches, successes and failures, will complement substantial industry investment in the research and development of needed technologies.

In response to interest expressed by the administration and as part of a continuing public sector/private sector dialogue, this report outlines consensus recommendations by leading academic and industrial researchers for a federal research and development agenda for the NII. Section B of this summary identifies the broad technical challenges that must be addressed by the agenda. Section C summarizes the recommended research and development agenda related to each broad technical challenge, and Section D provides the rationale for government investment in the research agenda. The remainder of the report provides background information on how the consensus recommendations were formulated, further details on the broad opportunities and technical challenges, and a more complete description of the recommended research agenda.

B. Technical Challenges for the NII

End-user applications in health care, education, manufacturing, commerce, finance and government dictate much of the functionality and performance required of the NII. Scenarios for these applications include remote patient-doctor and doctor-doctor consultations; access to remote diagnostic data during routine and emergency medical treatment; student access to educators, libraries and information repositories across the nation; distributed cooperative learning environments; and electronic commerce and financial transactions by businesses as well as individual consumers. From scenarios such as these, the following broad technical challenges are identified for developing an information infrastructure with the requisite functionality and performance:

Network components and protocols must be able to handle voice, video and text simultaneously, must interoperate seamlessly and efficiently, and must include wireless and wired communication. "Face-to-face" contact between patient and physician via the network can offer specialized medical resources to all citizens. In manufacturing, interactive capabilities that include moving large amounts of data at very high speeds will support geographically dispersed design teams, as well as on-line design changes to address production problems. Financial services will include interactive assistance and customization of products, as well as very high-volume, distributed real-time transaction processing.
Information appliances and servers will provide access and services in the NII. These system components must communicate efficiently and effectively with the user, and with networks and mass storage devices in the infrastructure. They also need to be scalable (to meet the needs of a diversity of applications and services), highly dependable and able to interoperate with all manner of NII services. High-volume transaction processing and very high-speed simulations of multiple futures for financial services are among the applications that require high-performance servers.
Information access techniques must enable efficient searches of large distributed information repositories and make a myriad of information resources understandable. Federal, state and local governments have vast amounts of information, distributed across many agencies and organizations, that must be accessible via the infrastructure. A municipality desiring to develop a recycling program should be able to obtain information on other municipalities and organizations that have implemented recycling programs for similar populations.
Multimedia information technologies, for example, the synchronized and integrated real-time delivery of voice and video, or search and retrieval based on image content rather than textual attributes, must be integrated into information access services. Patient records including X-rays, EKGs, lab reports, doctors' notes and video clips must be rapidly retrieved in emergency situations. Search on image content will greatly facilitate diagnostic procedures.
Infrastructure for applications development must provide common solutions that create markets and foster competition by enabling the construction of affordable, interoperable, easy-to-use applications that have an appropriate level of security and dependability.
Dependability and manageability must be designed into the NII from the outset and must allow for flexible growth with increasing and changing demands of NII users and the availability of increasingly complex services. Patients will require dependable communication with their doctors during remote treatment, and doctors will require dependable access to diagnostic data in emergencies. Furthermore, for electronic commerce the cost of NII downtime will be enormous.
Ease of use. The services provided by the information infrastructure must be accessible by users with widely varying skills, experiences, abilities and backgrounds. The interface must be easy to use by a student in the classroom, a physician in a clinic or a worker on the shop floor. In the health care arena, consultations between physicians must happen as effortlessly as ad hoc "hallway conversations." Distributed cooperative learning services also need to provide the same conducive environments as today's face-to-face environments.
Interoperability among heterogeneous systems is required on an unprecedented scale. Students will want to access libraries across the nation to obtain information for a term paper. Doctors, laboratories and insurance companies will require access to a patient's records. Interoperability must be provided between interconnected networks, between appliances and networks, between services, and between appliances and services.
Security and privacy technologies must be easy to use and must provide appropriate levels of security to suit the requirements, cost constraints and convenience of the end user. Merchants will use the NII for commerce by advertising and selling goods over the network. They will need assurance that NII payments are not fraudulent and will want sales figures and other business information to be protected from competitors. Financial transactions will require the use of "digital currency" that provides anonymous transactions with the same ease of use and flexibility as cash has today. The buyer requires privacy assurances, yet the seller must be assured that the payment is valid.
Portability, mobility and ubiquity are special technical challenges for the information infrastructure. Teachers will travel among several schools, financial traders may participate in transactions from various customer sites, and emergency medical personnel will treat accident and disaster victims en route to the hospital. The NII utility must be as widely available as today's telephone services. Users must be able to move from site to site and still access the NII at either end point or while in transit.

C. Recommendations

This section summarizes the recommended research and development agenda to address each of the above challenges. The overarching recommendation for pilot projects and testbeds is discussed first, followed by a summary of the further recommendations in each topic area.

Research in networking and distributed information technologies during the past 25 years has been guided by pilot projects on testbeds such as the Internet. Similarly, NII research should be guided by pilot projects to develop appropriate technologies, evaluate the technologies and get the technologies into the hands of the users. For example, the creation and experimentation with prototype NII systems that contain security and commerce mechanisms will catalyze the widespread deployment and use of these mechanisms. In the area of multimedia technologies, exploration of both education and health care testbeds will provide important diversity. Testbeds for applications development will serve to develop and validate the needed software technology, databases and innovative use of information. Interoperability paradigms must be identified and tested in realistic settings. In the domain of ease of use, existing testbeds such as Mosaic and NSF's Scientific Collaboratories should be evaluated to understand the properties that are required to facilitate individual use, social use and collaboration. Further testbeds should be developed to evaluate future walk-up-and-learn and collaborative system technologies.

The subsections below summarize the principal challenges and the additional research and development recommendations in each of the 10 technical topics considered.

C.1 Network Components and Protocols

The key challenge for the creation of network components and protocols for the NII is to create a single communications infrastructure that embodies and integrates the attributes that have made the separate telephone, television and computer communications networks successful. Because, like the Internet, the NII will comprise a multiplicity of separately administered networks, a key aspect of the above challenge is to evolve the protocol hierarchy that supports internetworking in the Internet to embrace the full range of applications and the full range of communications services envisaged for the NII. The Internet protocol hierarchy can be described as having an "hourglass" shape, in which a single common protocol that forms the "waist" of the hierarchy supports a variety of application and transport layers above and allows a diversity of physical layer transmission and switching services below.

The recommended research and development agenda to address the above challenge spans the fields of material science, electrical and computer engineering, networking and performance. The agenda includes the following topics:

Enabling technologies, including materials that underlie electronic, optical and opto-electronic components; all-optical networking technologies; powering technologies; coding and modulation technologies; survivability and reliability technologies; low-cost connector technology; and design tools for high-speed integrated circuits.
Design and implementation of low-cost premises networking technologies, low-cost access network technologies, high-speed backbone technologies, high-speed wireless technologies, and high-speed/high-capacity switching technologies that allow the separation of fabric and control and that permit effective management of capacity, including the provision of service quality guarantees.
Assuming the hourglass-shaped protocol hierarchy for the NII, define what functionality is needed in the "waist," and what techniques are needed for implementing these functions.
Investigate the trade-offs between a single NII protocol hourglass and multiple NII protocol stacks in terms of mechanisms for internetworking, cost efficiencies, flexibility and maintainability.
Analysis, based on network models, of future integrated services networks. In particular, the consequences of diversity in quality of service on network design and operation must be evaluated.

C.2 Information Appliances and Servers

"Computer" is an inadequate word to describe the client and server systems that connect people to the NII and provide its services. These systems must be "information appliances" and "information servers" that support communication, information storage and user interactions--and incidentally compute. The key challenges in this area are that NII systems must:

1) Be communication systems. Both the hardware architecture and the software architecture must tightly integrate communication.

2) Support "plug-and-play" networking. Appliances must be compatible at any site at many levels of abstraction.

3) Accommodate extraordinary diversity in protocols, functionality and services.

4) Be highly scalable in cost and performance.

5) Be highly reliable and available--at least as reliable and available as today's telephone system.

6) Be "future-proof." The servers on the NII will represent a major investment that will be upgraded over time, and therefore these systems must be designed to be evolvable.

The recommended research and development agenda to address some of these challenges spans the fields of computer engineering, computer architecture, distributed systems, software engineering and performance modeling. Key recommendations include:

Performance analysis: Research should advance understanding through analysis (modeling and simulation) and synthesis (prototyping) of information appliances and services.
Network interfaces and memory systems: Moving network interfaces closer to high-bandwidth processing (and ultimately onto commodity microprocessors) to both improve performance and lower cost.
Storage hierarchies: Less expensive storage hierarchies to meet new (and often conflicting) demands--larger, faster and non-volatile.
Resource management: Programming models and application programmer interfaces that support application/system negotiation and feedback.
Distributed systems architecture: Techniques for building reliable services from collections of servers on a scale never before considered.
Special-purpose support: For encryption/decryption, compression/decompression, speech recognition/synthesis, video recognition/synthesis and interactive multimedia.
Advances in technology: Inexpensive displays that strive for the resolution of paper.

C.3 Information Access

There are four key challenges for the creation of an NII information access utility:

1) The creation of effective user interfaces that tolerate imprecise requests and make a myriad of information resources understandable.

2) The creation of highly efficient, yet effective, methods for searching distributed information repositories to obtain answers to specific queries.

3) The creation of large-scale information resources that assist users in understanding their content.

4) The creation of architectures that enable the integration and interoperation of separately designed information resources.

The recommended research and development agenda to address some of these challenges spans the fields of databases, artificial intelligence, human factors and performance evaluation. Key recommendations include:

Advanced query languages and capabilities that provide location-independent access, anticipate users' needs and assist the user in formulating precise requests.
Interfaces capable of adapting to user preferences, limitations and behavior.
Mechanisms for effective information filtering.
Sophisticated techniques for optimizing searches of large, mobile and distributed information resources, including capabilities for time-critical delivery of data and for handling conflicting responses or out-of-date data.
Information repositories that support the above capabilities.
Translation methodologies, content languages and the development of application-specific ontologies that facilitate the integration and interoperation of multiple information resources.

C.4 Multimedia Information Technologies

There are three key challenges for integrating multimedia capabilities in the NII information access services:

1) The requirement for simultaneous, integrated and real-time delivery of multiple media streams.

2) The requirements for more-complex operations such as search based on shapes and/or hues, which present computational and user interface challenges.

3) The consumer market requirements for low cost, ease of use and large scale (i.e., large number of information sources and users).

The recommended research and development agenda to address some of these challenges spans the fields of device engineering, databases, artificial intelligence (vision) and economic modeling. Key recommendations include:

Information capture and generation, including automatic feature abstraction and indexing for retrieval.
Representation of multimedia objects, including modeling of the relationships and coordination between multiple streams, and semantics-based compression schemes.
Information management systems that deliver multiple data streams in an integrated fashion with guaranteed time constraints and support content-based queries.
Ownership and fair-use issues, including economic and legal models for multimedia in the NII and media registration facilities to protect intellectual property.
Low-power storage hardware for mobile multimedia access devices.

C.5 Infrastructure for Applications

The key challenge is to create an infrastructure to support low-cost and rapid development of a myriad of applications. Specific challenges include:

1) The requirement for interoperability.

2) The need to leverage common solutions for security, dependability, ease of use and distribution over heterogeneous systems.

The recommended approach is to provide functionality in the form of reusable objects that provide services. These objects can be used as software components by application developers. The recommended research and development agenda involves refining this approach and researching the requisite technologies, which span the fields of software engineering and distributed computing. Key components of the agenda include:

Studies on test applications, such as financial services and health care, to refine the model of building NII capabilities from coarse-grain objects or services.
Development of object system foundations including:

a) Primitives for replication and caching of objects while maintaining consistency and managing persistent storage.

b) New (hybrid) models for communication and coordination among the distributed components of an application.

c) Primitives for collaborative applications.

d) Server load-balancing brokers.

e) Visualization and performance tuning tools.

f) Application construction tools.

C.6 Dependability and Manageability

The word "infrastructure" connotes dependability to many users. Six key challenges for developing a dependable and manageable NII are:

1) Dependability and manageability must be built into the system from the start--they cannot be afterthoughts.

2) Distributed management of the NII must be coordinated so that one party cannot optimize operation at the expense of another.

3) Quantum increases in scale and service will stress old solutions.

4) Precautions need to be taken to ensure that new applications and services can successfully coexist with the existing infrastructure (e.g., progress must be balanced with stability).

5) Measuring communications and information services so that failure data is trapped for analysis is crucial, but how to do this is poorly understood.

6) Mechanisms for dealing with perennial problems such as network and server failures or overload, security violations, incompatibility of interoperating systems, and clashes and inconsistencies between global and local policies are needed.

The recommended research and development agenda to address some of these challenges spans the fields of networking, reliability and system modeling. Key recommendations include:

Continuous research in scaling up, refining and improving current solutions.
Techniques, metrics and benchmark suites for characterizing and validating service quality, and monitoring infrastructure performance.
Techniques for improving the reliability and availability of NII system operation, such as configuration management techniques that enable non-disruptive introduction and reconfiguration of services; replication strategies for masking failures; fallback mechanisms and graceful degradation; resource optimization and security administration; and reduction and visualization of system management data.
Design and development methodologies for orderly growth and evolution, including simulation and modeling technologies for evaluating product performance and interaction.

C.7 Ease of Use

There are four key challenges for developing ease-of-use technologies:

1) Universal access implies a very diverse population of users.

2) Interfaces must provide adequate access on inexpensive devices.

3) Interfaces should ideally provide all users with equivalent access to information despite differences in their skills, experience and cognitive capabilities.

4) Effective interfaces and interaction paradigms are needed for a variety of information appliances.

The recommended research and development agenda to address these issues spans the areas of psychology, sociology and user interfaces. Key recommendations include:

In-depth psychological, social and cultural study of NII users to:

a) Understand how individuals can use their previous real-world experience to navigate through the NII using walk-up-and-learn interfaces.

b) Identify opportunities for improving the quality of life at work and home.

c) Understand how to support various levels of multiuser collaboration.

Development of technologies that promote ease of use, such as low-cost, high-performance (portable) hardware displays, adaptive intelligent interfaces, human-centered interfaces such as pen and speech, user interface standards and advanced visualization techniques.
Metrics and methods for evaluating the ease of use of NII applications and services.
Development of a core set of user interface standards to enable easy-to-use services across different platforms.

C.8 Interoperability

The key challenges to creating interoperable communications and information services in the NII are:

1) The specifications that form the basis for designing and implementing large distributed software and hardware systems are inherently imperfect, leading to syntactic and/or semantic mismatches.

2) Integrating legacy and other systems that are designed and implemented completely independently. Opportunities that might help meet these challenges are the development of open systems and the development of reusable software technology.

One goal of research and development efforts in interoperability is to enhance our ability to compose systems into a single acceptably seamless and efficient system. Another goal is to develop an intellectual framework for discussing, quantifying and demonstrating interoperability. Within the framework of these broad goals, some of the key recommendations include:

Identify, compare and contrast interoperability paradigms on realistic problems in various interoperability contexts. Generic as well as application-specific paradigms should be studied.
Identify and define common generic and application-specific services that provide a range of functionality (e.g., minimal to maximal), and develop mechanisms for making these services widely available in the "NII operating system."
For various interoperability contexts, identify key protocols and interfaces that might be widely adopted.
Develop methods that can predict during the design phase how well the composition of constituent systems will interoperate.
Develop tools and testbeds that can be employed during the integration phase to test whether each system to be composed conforms to the protocols and interfaces it uses.

C.9 Security and Privacy Technologies

The two key challenges for developing needed security and privacy technologies are:

1) Providing an adequate level of base security, to reduce the need for security impediments to interoperability.

2) Providing an adequate level of security for electronic commerce, for which the requirements are not well understood.

Further challenges include providing the capability to enhance or reduce security to suit the requirements, cost constraints and/or convenience of the end user; developing security mechanisms that are easy to use; and developing techniques that operate across national borders.

The research and development recommendations for security and privacy technologies that will address these challenges include:

Creation and experimentation with prototype NII systems that contain security and commerce mechanisms.
Social and organizational research to understand how to build and maintain organizations to provide system-level security.
Research in human engineering to define security mechanisms that are sufficiently unobtrusive that users will not avoid or ignore them.
Research in a variety of systemwide security issues including:

a) Integration and testing of security mechanisms in environments that pose real threats.

b) Development of a well-specified security architecture that facilitates both mandatory and discretionary security, risk assessment and extensions as NII services and capabilities expand.

c) Mechanisms for security management across network and national boundaries.

d) Technical and social mechanisms for security assurance.

e) Policies that are easy to use by non-expert users.

f) Development of threat models that draw on published experience with actual threats.

g) Mechanisms for detecting and preventing fraud without violating the privacy of honest users.

Research in core security services, such as authentication, encryption, authorization and anonymous access, that scale to billions of users and/or terabit data rates.
Research to identify and address the requirements of applications such as electronic payment systems, electronic contracting and electronic commerce.

C.10 Portability, Mobility and Ubiquity

There are six key challenges for portability, mobility and ubiquity:

1) Wired and wireless networks with inexpensive connections.

2) Interoperation between wired and wireless networks.

3) Universal connectivity for equipment with differing protocols, bandwidth, frequency and capability.

4) Quality-of-service guarantees, including real-time guarantees for continuous media streams, in a dynamic environment where users move and channel conditions change.

5) Interoperation of applications and services under intermittent or degraded connectivity.

6) Security mechanisms for an environment in which no guarantees about physical integrity can be provided.

Key research and development recommendations to address these challenges include:

Explore optimal trade-offs in coding, power, bandwidth and cell size in microcells.
Portable terminal componentry.
Wireless components and systems technology, including protocols for mobile networks.
Universal connectivity, including concurrent communications and efficient information dissemination methods, as well as a "universal bootstrap mechanism" (i.e., download from the NII software needed to interact with the NII).
Robust security systems and compression technology for wireless transmissions.
Replicated data management methods and programming paradigms for intermittently disconnected environments.
Low cost, non-volatile storage with fast access, high data rates and low power requirements for geographically dispersed and mobile databases.

D. Rationale for Government Investment

Research and development efforts that lead to NII technologies will advance the public welfare and economic goals as discussed in Section A. Much of the investment in the NII is coming from the various private-sector organizations that are and will be leading the development and deployment of NII products and services. Many of these organizations will sponsor and carry out substantial research and development efforts in the underlying technologies for the infrastructure. For maximum effectiveness, the federal research and development agenda must complement these commercial research and development efforts in a way that stimulates and catalyzes further investment from industry. The agenda recommended in this report does this by focusing on 1) long-term and high-risk challenges and 2) shared technologies as discussed below.

Attention to the research agenda recommended in this report can be an important stimulus to commercial investment, in part because many of the research problems involve greater risk and/or longer horizons than are typically feasible for commercial research and development. For example, sophisticated multimedia search and retrieval technologies will have great value in medical and other applications, but there is not a clear way to achieve the goals. As another example, exploratory research in ease-of-use and/or information appliance technologies may create new visions of how users will interact with the information infrastructure. Government has an essential role in investing in this kind of long-term, high-risk problem-solving and exploratory research to create new concepts and technologies that are then refined through competitive commercial research and development efforts into new-generation products and services. The recommended demonstrations will serve as testbeds for assessing integration and scaling issues and provide mechanisms for "bootstrapping" the introduction of major new NII technologies and facilitating the emergence of new markets.

Just as importantly, the creation of the NII will require or derive great benefit from the shared use and broad dissemination of results for many of the research and development problems recommended in this report. For example, research and development of common interfaces and common services will enhance the interoperability and rapid development of the information infrastructure and applications. Common system interfaces help application developers decouple their design choices, thus making it easier for developers to respond to new opportunities and for suppliers to enhance their offerings. Common services and broad dissemination of research results in areas such as ease of use, dependability and manageability, privacy and security, and information access enable the rapid development of applications and information repositories that can use the results rather than reinventing the solutions. Furthermore, large-scale testbeds will have great value for introducing and understanding prototype technologies if the results are shared. As public assets, these results and standards will promote a competitive response to the needs of major public-interest applications of the NII such as health care, education, and government information and services.

In summary, the research and development agenda in this report will create 1) new concepts on which future markets may be based, 2) commonalities that facilitate effective and competitive response to the needs of the major users of the NII and 3) useful testbeds that can introduce major new technologies. By fostering such efforts, government will continue an effective partnership with industry and academia that has enabled the remarkable progress in information technology over the past three decades and will drive forward the needed NII technologies--as well as the markets they create.