Concepts & Innovation

Here we present interesting topical technology and introduce concepts and ideas which could well shape the future.

What would you prefer - High Altitude Platforms or geostationary satellites ?

Pico-satellites with Micro-ElectroMechanical Systems (MEMS)

Internet Protocol (IP) - back to future ?

Transmission Control Protocol (TCP) problems over satellite

Soft computing - emulating humans ...

Brain-computer interface .. read on

High Altitude Platform (HAP) or High Altitude Long Endurance Platform (HALEP) are stratospheric platforms being investigated for broad-band wireless telecommunications. The platform are hoisted at altitudes of about 17~22 Kms to set communication links similar to satellites. Within this height range, wind currents are low and commercial aviation is unaffected.

So why the interest ? The answer is much simpler than the technology - there is a considerable demand for broad-band wireless communications and HAPs are a promising medium. They appear at a high elevation angle compared to terrestrial base stations thereby mitigating the terrestrial propagation effects, and with a visibility of around 200 Km at 5 Deg elevation, they can replace a large number of terrestrial base stations; and being considerably closer to the ground than satellites offer much lower path loss than satellites - better by ~ 34 dB for LEO satellites and ~ 66 dB for GEO satellites. HAPs have been assigned frequency bands in 47/48 GHz and 28 GHz (ITU region 3 only), bands where at present spare spectrum is in plenty. HAP telecommunication systems can be designed to respond dynamically to traffic demands; they are relatively low cost compared to satellites (perhaps, $50 million ? ); they can be deployed incrementally and rapidly when necessary; the platforms and payload are upgradeable; and they are environmentally friendly using solar power, without need of launchers and eliminate the need of terrestrial masts. Typical cell size of HAPs range between 1-10 Km and the communication throughput can range between 25-155 Mb/s. The coverage is regional, though it is possible to inter-link HAPs creating a national grid, or alternatively they can be connected to distant gateways via satellites. They have been proposed for broad-band fixed wireless access (B-FWA), mobile communications as base stations, rural telephony, broadcasting, emergency/disaster applications, military communications, etc.

With such advantages why have they not been exploited commercially? Well, the technology has yet to arrive. There are a number of open technical issues being actively pursued. A number of system issues are under investigation including - system architecture, frequency planning, network protocols, resource planning, etc. Propagation characteristics at 47/48 GHz are not well defined yet; modulation/coding techniques have to be optimized for such propagation conditions; 48 GHz antenna technology with multiple spot beams is under development. Other issues include platform station-keeping, hand-off considerations even for fixed stations due to platform movement and payload power. HAPs have similar eclipse problem to satellite with regards to payload power due to the use of solar cells. Reliable platforms are yet to be developed. Platforms under investigations include Airship , Airplane, unmanned aerial vehicle and tethered aerostat which reach up to 5 Km. Airships use very large semi-rigid helium-filled containers. One implementation of airplane HAP is the solar powered plane which flies in a tight approximate circular path.

A number of research and commercial programs are addressing the problems. They include HeliNet Program, HALO Project, SkyNet and others (see reference) . The technology is a few years away from implementation (2001) but entrepreneur are ready to launch their system today only if the financiers were a bit less condescending. To be impartial - their innocent question is - when could we expect a return ? It is believed that HAPs will arrive when the technology has matured to make them commercially viable.

Due to overall costs and complexity in providing world wide seamless coverage it is unlikely that HAP will replace satellite, but synergistic solutions can, in fact, augment traffic carried by each. Both the books of the author refer to the concept of communication through HAPs in context of their role in a satellite communication system.


1. Tozer T. C, and Grace D (2001). 'High-altitude platforms for wireless communications', Electronics Communication Engineering Journal, June, pp. 127-137.

2. Djuknic G. M, Freidenfelds J. and Okunev Y. (1997). 'Establishing wireless communications via high altitude aeronautical platforms: a concept whose time has come ?' IEEE communications magazine, September, pp.128-135.

3. Helinet

4. Advanced Technologies Group

5. SkyStation


Micro-electromechanical systems or MEMS as the name implies are tiny systems which combine semiconductor technology with micro mechanical devices such as gears, diaphragms, fluid thrusters, accelerometer, motors and heat controllers built typically on silicon thus giving the aerospace industry the possibility of creating very small or 'pico' satellites weighing less than a kilogram.. MEMS are already in use as car airbag triggers, low power filters on mobile phones and in optical fibre routers as switch. MEMS satellites when mass produced could be used for a vast array of satellite-based applications with advantages in terms of low launch costs, high resistance to radiation and vibration (due to their micro size).

A number of MEMS devices are under development - switches, gyros, thrusters, thermal control systems, propulsion system in institutions like the Jet Propulsion Laboratory (JPL), Pasadena, and Aerospace Corporation in El Segundo. Several MEMS satellites have been flown for testing components. The first application of a MEM satellite may be a tiny satellite inspector which will be used to inspect faults housed within a mother spacecraft to fly around it to inspect in case of malfunction thereby saving considerable diagnostic efforts on the ground. It is anticipated that the technology is at least one to two decades away.


1. Crass S (2001). 'MEMS in space', IEEE Spectrum, July, pp.56-61.

2. IEEE Spectrum (1998). 'How to model and simulate microgyroscopic systems', June, pp. 66-75.

3. IEEE Spectrum (2001).'Large jobs for little devices', January pp. 72-73.


Internet Protocol suite or IP was developed in 1970s as a part of research funded by DARPA (Defence Research Projects Agency) for the experimental network ARPANET and in 2001 there is evidence that the protocol will form a backbone transport protocol of future telecommunication networks. It is a remarkable achievement considering the unprecedented changes in telecommunications technology in recent years.

Why is IP so resilient ? In a nutshell, the protocol is general and extendable.

IP is an end to end protocol which operates without specific knowledge of the network and therefore passes on the error recovery to the end system as it does not expect cent percent network reliability. This implies that protocol can traverse a mix of networks during transport - e.g. Ethernet, X.25, ISDN, satellite network, etc.

IP supports two types of end-to-end transport services - TCP, which has features to reduce network congestion and errors using a handshake between end services and UDP, which is a basic transport mechanism without the handshake feature of TCP. This way IP can support a mix of services such as video, voice, data streaming.

The 32 bit address field in conjunction with a decision to maintain a global address registry, allowing each IP network to keep its own address repertoire is a key to the extension of the Internet as a global network.

The IP protocol suite is modular, thereby allowing independent enhancements to each component of the stack. For example numerous modifications to the routing protocols have allowed scaling up the number of computers without affecting any other layer. Similarly, TCP suite has been continually refined over the years to improve network congestion and user feedback mechanism without affecting other layers while retaining backward compatibility with earlier TCP generations.

Acceptability of IP protocol has been aided by the ready availability of the protocol specifications and reference implementation; openness in its formulation; and changes to the suite with a general consensus - a model which continues to be practice in the Internet Engineering Task Force (IETF).

However, despite the universal acceptability of the protocol, a number of improvements are necessary to usher the paradigm to the next generation Internet and other telecommunication networks which intend to incorporate IP as a transport backbone.

Routing and traffic management continue to present challenges as Internet growth continues and there appears to be a need for better management of traffic to avoid traffic congestion by adaptive routing. IP operates with the notion of identity and location within its address which is not amiable to mobile communications. Identity and hence authentication aspects are inadequate, due to its support of a distributed architecture based on trust. The suite is not amicable for multicast, an application which is growing in demand. Another area in need of enhancement is the support of quality of service despite several alternative and often propriety solutions. TCP mechanism within the suite is inefficient in wireless environment, and in particular, on satellite routes. The 32 bit address space of IP suite is already beginning to reach its limits.

IPv6 has been developed to improve several IPv4 features including the address limitations by increasing the address space to 128. However, its introduction to the Internet is slow. In the meanwhile users continue to introduce interim solutions to skirt existing limitations.

A note addressing TCP limitations over satellites is available in 'Innovations'.


1. Huston G, 'The IP scorecard', (2001). Satellite broad-band, August. pp. 20-22.

2. Richharia M, ' Mobile satellite communications, Principles and trends', (2001). Addison and Wesley, pp. 529-535.



Transmission Control Protocol (TCP) is a commonly used transport protocol for non real-time application over IP networks. The protocol was developed for fixed links assuming that there are negligible errors or transmission delay in the link and packets are lost only due to congestion. The assumptions do not apply to satellite links with long delays and transmission errors therefore TCP/IP throughput over satellite channels can degrade. Problems are caused by a number of reasons : (a) slow start up of TCP protocol; (b) transmission errors; (c) link asymmetry; (d) attempts to transfer real-time data over TCP.

Slow start up problem occurs because TCP ramps up the transmission rate gradually while assessing network capacity. The capacity is probed by awaiting acknowledgment of received packets. The congestion window is initially set to 1 and increased on each successful receipt of acknowledgment. In geostationary, medium earth orbit and low earth orbit systems, operating at 1 Mb/s the full capacity can be reached respectively in about 3.91, 1.49 and 0.18 seconds when using 1 Kbyte packets.

A number of solutions are used to circumvent the slow start problem. Common solutions include - increasing the initial congestion window, TCP spoofing, cascading TCP connections, fast start and sudden start. Increasing initial window size can give an improvement of the order of 3 times the round trip delay. In TCP spoofing technique, an artificial acknowledgment is sent back to the source by a router near the sender, while the satellite link is established at the far end, thus the TCP ramps up throughput rapidly. In fast start TCP transmission rate is derived from recent history. In sudden start algorithm, dummy packets are sent initially to gauge the network conditions. At the end of this training phase the transmission rate is ramped up to the level acceptable by the network.

Transmission errors compound the slow start problem, as the packets lost due to errors are assumed lost due to congestion and therefore throughput is throttled back. In addition, the lost packets have to be retransmitted.

Proposed solutions include Explicit Congestion Notification (ECN), Link Corrupt Notification (LCN) and Rapid Recovery. In ECN scheme, congestion is notified to the sender; and packets not lost by congestion are assumed lost due to link errors. Transmission rate is not decreased on encountering transmission losses. In LCN, the receiving station notifies bad link conditions to the sender by continually monitoring the link and the sender takes appropriate measures depending on channel conditions.

Satellite links are asymmetrical in a number of applications with a low throughput in the return direction. This causes traffic burstiness and loss of throughput by TCP whenever the return link is congested.

Proposed solutions to the bandwidth asymmetry problem include header compression methods, periodic acknowledgment by the receiver, reduction in acknowledge data flow through a feedback, ACK filter, sender adaptation, ACK reconstruction and satellite transport protocol (STP).

None of the solutions are ideal and therefor considerable development effort is underway in industry and research institutions.


1. Allman M et al,'Ongoing TCP research related to satellites' (2000). RFC 2760, February.

2. Akyildiz I.F, Morabito G and Palazzo S, 'Research issues for transport protocols in satellite IP networks' (2001). IEEE Personal Communications, Vol 8, No 3, June, PP 44-48.


Soft computing in satellite arena

Imagine having to operate a mobile satellite communications network where traffic is changing unpredictably, users are constantly demanding attention for a variety of reasons, gateways are malfunctioning due to - scintillation, rain fade, software crash ....... This type of nightmarish operational scenario, transparent to the users, is routine. In real world, things are not as precise as machines would like and therefore humans must keep a vigilant eye on state of affairs as they have the experience and the skill to react to unforeseen and unexpected events. Conventional 'hard' computing follows rules and dogma laid out by its creator whereas high IQ machines using soft computing have an ability to manage 'imprecise' problems.

Soft Computing (SC) is an emerging discipline attempting to provide robustness in presence of uncertainty, imprecision and partial truth of real world by synergistically combining an array of computing technologies - fuzzy logic, neurocomputing, probabilistic computing, chaotic computing and machine learning. SC algorithms by contrast to conventional or 'hard' computing are able to advantageously exploit natural phenomena such as intuition and subjectivity, allowing modeling ambiguity and uncertainty as a human would.

SC methodologies are being used in a number of industries being particularly useful when traditional analytical methods fail. Intelligent systems can be constructed through flexible knowledge acquisition and processing in conjunction with powerful knowledge representation. SC has been applied to in aerospace, communication systems, consumer appliances, electric power systems, manufacturing automation, robotics, power electronics and motion control, processing engineering, transportation, etc..

SC have been used or proposed for - orbital operations of Space shuttle, aerospace and aircraft control systems, channel assignment in cellular and satellite systems, multimedia traffic prediction, design of network topologies, processing of code division multiple access, etc.



1. Proceedings of IEEE, September 2001, Special issue 'Industrial innovations using soft computing'.


The electrical activity of human brain captured in eletro-encephalograph (EEG) changes in response to human thoughts. EEG based brain-computer interface make use of thought signals to trigger an appropriate action in a computer. While the technique is being researched for medical applications - for example, to aid a paralysed patient, its application can be conceived for other industrial application. An astronaut could exercise such an interface when repairing a space vehicle, a motorist could dial a telephone number or tune the radio channel, etc..

In reference 1 the authors demonstrate an EEG-based computer interface for restoring the hand grasp function of a paralysed patient with an electronic hand device using mental imagination of the motor command and reference 2 reports a number of related research papers.



1. Pfurtscheller G and Neuper C 'Motor imagery and direct brain-computer communication' Proceedings of IEEE, July 2001, PP 1123-1134.

2. Proceedings IEEE, July 2001, 'Neural Engineering: Merging engineering & Neuroscience' .