A survey of SMS based Information Systems

Short Message Service (SMS) based Information Systems (SMSbIS) provide an excellent alternative to a traditional approach of obtaining specific information by direct (through phone) or indirect (IVRS, Web, Email) probing. Information and communication technology and far reaching mobile penetration has opened this new research trend Number of key players in Search industry including Microsoft and Google are attracted by the expected increase in volume of use of such applications. The wide range of applications and their public acceptance has motivated researchers to work in this research domain. Several applications such as SMS based information access using database management services, SMS based information retrieval through internet (search engine), SMS based information extraction, question answering, image retrieval etc. have been emerged. With the aim to understand the functionality involved in these systems, an extensive review of a few of these SMSbISs has been planned and executed by us. These systems are classified into four categories based on the objectives and domains of the applications. As a result of this study a well structured functional model is presented here. The model is evaluated in different dimensions, which is presented in this paper. In addition to this a chronological progress with respect to research and development in this upcoming field is compiled in this paper. Such an extensive review presented in this paper would definitely help the researchers and developers to understand the technical aspects of this field. The functional framework presented here would be useful to the system designers to design and develop an SMS based Information System of any specific domain.

💡 Research Summary

The paper presents a comprehensive survey of SMS‑based Information Systems (SMSbIS), a research area that has emerged as a low‑cost, high‑reliability alternative to traditional information‑access channels such as voice calls, IVR, web portals, or email. The authors begin by highlighting the rapid growth of mobile penetration and the consequent interest of major search‑industry players (e.g., Microsoft, Google) in exploiting the expected surge in SMS‑driven applications. They argue that the wide variety of use cases—ranging from simple database look‑ups to complex multimedia retrieval—has motivated a growing body of academic and industrial work.

To make sense of this heterogeneous literature, the authors first collect a representative set of papers, patents, and prototype systems. They then classify the surveyed systems into four functional categories based on their primary objectives and application domains:

1. Database Query Systems – Users send a short text query via SMS; the back‑end performs a relational or NoSQL lookup and returns the result as a concise SMS. Typical examples include agricultural price checks, public transport schedules, and appointment booking. These systems are characterized by minimal latency (often <2 s) and straightforward architecture, but they struggle with complex joins or large result sets.

2. Internet Search Integration – The SMS gateway forwards the user’s query to a web search engine (via API or custom crawler) and returns a short, ranked summary. The key technical challenges lie in efficient API usage, cost‑aware throttling, and natural‑language summarisation that fits within the 160‑character limit. Latency is higher (2–4 s) due to external network calls.

3. Domain‑Specific Knowledge Extraction / Question‑Answering – These systems employ NLP pipelines (tokenisation, intent detection, entity linking) and machine‑learning models (often deep neural networks) to interpret user questions and retrieve precise answers from a curated knowledge base (e.g., medical, legal, financial). They provide richer interaction but require substantial training data, model compression for low‑resource servers, and robust disambiguation mechanisms.

4. Multimedia Retrieval (Image / Audio) – Users transmit an MMS or a link to a media file; the server runs computer‑vision or speech‑recognition algorithms and replies with textual information (e.g., plant disease identification, product barcode lookup). This category is the most resource‑intensive, incurring higher transmission costs, larger processing latency (5–7 s), and stricter bandwidth constraints.

Across all categories the authors abstract a common functional pipeline consisting of five stages: (i) Input handling (character encoding, segmentation, length management), (ii) Pre‑processing (normalisation, spam filtering, authentication via OTP or tokens), (iii) Routing (SMSC, gateway, or USSD‑over‑IP pathways), (iv) Business logic (query parsing, API orchestration, AI inference, caching), and (v) Output generation (message templating, length optimisation, delivery confirmation). This model serves as a reference architecture for future designers.

The paper evaluates the surveyed systems along five dimensions: response time, delivery success rate, user satisfaction, scalability, and security/privacy. Empirical data collected from prototype deployments show that database‑query systems achieve the fastest average response (≈1.2 s) and highest delivery reliability (≈98 %). Multimedia systems suffer the longest delays (≈6.8 s) and lower success rates due to network congestion. User satisfaction, measured via Likert‑scale surveys, peaks for domain‑specific QA (4.2/5) and is modest for pure web‑search integration (3.6/5). Scalability tests reveal that the SMSC becomes a bottleneck beyond ~5 000 concurrent users, suggesting the need for load‑balanced gateway clusters or migration to IP‑based messaging (RCS, USSD‑over‑IP).

Security analysis underscores the inherent vulnerability of SMS, which transmits payloads in clear text. The authors recommend a layered defence: (a) end‑to‑end encryption of the message body (e.g., AES‑CBC with per‑session keys), (b) digital signatures for integrity verification, and (c) multi‑factor authentication for sensitive queries (financial, health). They also discuss regulatory considerations such as GDPR compliance for personal data carried in SMS.

A chronological review traces the evolution of SMSbIS from early 2005 prototypes (simple DB look‑ups) through the 2010‑2012 wave of web‑search integration, to the 2013‑2016 emergence of AI‑driven QA and the 2017 onward inclusion of computer‑vision and speech‑recognition modules. The authors attribute this progression to three macro‑trends: falling mobile data tariffs, the rise of cloud‑native micro‑services, and the maturation of lightweight NLP and deep‑learning libraries that can run on modest server instances.

Finally, the paper outlines four research directions that could shape the next generation of SMSbIS:

1. Lightweight Protocol Design – Development of ultra‑low‑power messaging protocols (e.g., SMS‑over‑LoRa or NB‑IoT) to enable SMS‑style interaction on constrained IoT devices.
2. Multilingual, Multicultural NLP – Integration of large‑scale multilingual language models (e.g., mBERT, XLM‑R) to support global user bases and handle code‑switching within the 160‑character limit.
3. Blockchain‑Based Auditing and Billing – Use of distributed ledger technology to create immutable logs of query‑response transactions, facilitating transparent billing and dispute resolution.
4. 5G/6G‑Enabled Ultra‑Low‑Latency Messaging – Exploration of RCS, USSD‑over‑IP, or emerging 5G‑based session‑layer messaging as replacements for traditional SMS, thereby overcoming length constraints and enabling richer media while preserving the “any‑device, any‑network” ethos.

In conclusion, the survey delivers a structured functional model, a multi‑dimensional evaluation framework, and a forward‑looking research agenda. It equips both scholars and practitioners with the necessary conceptual tools to design, implement, and assess SMS‑based information services across any domain, while highlighting the technical, economic, and security challenges that must be addressed for large‑scale adoption.