Jun Li

Memory Spot Application Framework

Memory spot is a microchip invented at HPL Bristol. The physical size of the chip is only about 2 mm². Through the Radio Frequency (RF) power coupling, the on-chip customized processor responds to the reader/writer over the RF channel on its embedded flash memory access. From a technology perspective, memory spot can be viewed as a Near Field Communication (NFC) device. However, compared to Radio Frequency Identification (RFID) and other NFC devices, memory spot not only occupies a much smaller physical dimension, but also provides a much larger storage capacity (up to 512KB) and a faster data transfer rate (~10Mb/s). Due to the limited processing power on the chip, applications have to be deployed on a host device, which can be a PC or a mobile device like a Personal Digital Assistant (PDA) or cellular phone. The reader/writer is attached to the host as an accessory device. Typically the application reads the data to the host cache through the reader/writer by positioning the reader/writer close to the spot (~ 1 mm away), manipulates the cached data at the host, and finally updates the data back to the spot. Because of its sufficient storage, applications can access the data on memory spot locally, anytime and anywhere, in contrast to other reference-only systems like RFID that have to rely on an online infrastructure for actual data access.

Memory spot can potentially become a significant enabler for a large collection of applications, including voice-annotated documents, electronic passport, medical history tracking, computer warranty tracking, etc. These applications span across both consumer and enterprise service domains.

We envisage memory spot as an integrated part of a much larger information processing system, which involves both the people that access the spot and the online infrastructure that the application occasionally interacts with. In enterprise service-oriented applications, such as to keep track of a computer’s configuration, service and warranty entitlement over its lifetime, or to record medical history information, memory spot allows data access such that different people from different organizations can read and update the data. Consequently, memory spot becomes a cross-organization data fusion point. Further, data access cannot rely on an always-on infrastructure, as otherwise the key value of self-contained local storage would be largely diminished.

For memory spot and its applications to be operated in a disconnected and cross-organization environment, our goal is to figure out how to provide a flexible programming environment to rapidly develop memory spot applications that can efficiently, reliably, and securely access and update the data stored in a memory spot. In particular, we have identified the following major technical challenges on data management:
• Flexible application data types: Data type definitions differ among applications, e.g. only an integer counter is required for a spot-enabled public transit pass, but complete computer service history requires an in-memory database (with different database tables) for efficient data inspection via SQL. A unified data modeling scheme is required to flexibly express arbitrarily complex data types;
• Reliable and efficient data access and update: The communication channel between the reader/writer and the chip is inherently unreliable, since the distance to the chip is often not well-controlled by the user. Sufficient feedback on the status of data access and update is required in order to assist the user to recover the communication channel promptly. Furthermore, flash-memory’s write access is relatively slow due to memory cells’ re-programming, which demands an efficient data update scheme different from a traditional file system;
• Scalable storage: For the applications involved with multimedia data, such as document scanning and voice annotation, the contents could exceed a single memory spot’s storage capacity;
• Data Security: data integrity, data confidentiality and data access control are crucial to high-value application data. However, one should not expect support from an always-on infrastructure to help ensure data security at the time of data access

To demonstrate the innovative applications enabled by memory spot and the application framework, we have prototyped the warranty spot application, which aims to significantly reduce computer-related warranty fraud, by having a memory spot permanently attached to each computer.

A memory spot is permanently attached to a computer and over its lifetime, machine configuration, service records and part warranty entitlements are all stored to the memory spot, by technicians or end-user customers or other authorized personals. Such computer information is ready for retrieval right next to the computer, without resorting to the centralized database, which does not exist today. Currently, such computer-related information is only available at the disjoint databases scattered across different organizations. To acquire comprehensive information about a computer under warranty is time consuming, or very often impossible as cross-organizational access is unavailable due to the issues in access authorization or internet connectivity, among others. As a result, the memory spot becomes a data fusion point of data provided from different organizations, regarding the computer that the spot is attached to.

The spot is called warranty spot, because it is originally designed to reduce warranty fraud for the computer under service contract. It can also be leveraged to improve service efficiency, as all the historical information on the computer is available, right next to the computer, independent of whether the computer is still functional or not.

The initial memory spot is created at the end of the manufacturing line, when all the major hardware components are configured. Once the computer is installed to the customer site and becomes operational, every time a technician from a different service provider is dispatched to provide the service, the technician can locally determine the most recent system configuration and whether the failure part is under warranty, by retrieving the information stored on the warranty spot, without relying on the online databases that might not have access rights granted to the external service providers, or at the time of the service, either an online connection is not available or the online service is temporally down. After the technician finishes the computer service, the service record is updated to the memory spot, along with the information about the hardware add-on parts and the warranty associated with the newly introduced parts. All data input will bear the provider’s digital signature to prevent data tampering and facilitate future data traceability.

Once the technician is back to the online environment (such as the office), the fully cached information from the memory spot can be used to provide data back up, in case the memory spot attached to the computer is maliciously destroyed or accidentally corrupted. The same information can be used to automatically fill a filed report. Furthermore, once this cached information is sent to the warranty administration department, the claimed part’s warranty entitlement can be verified, in contrast to the current practice that only relies on the part serial number to determine the warranty entitlement.